Codex compatibility note:
- Invoke repository skills with
$skill-name in Codex; this mirrored copy rewrites legacy Claude /skill-name references.
- Task tracker mandate: BEFORE executing any workflow or skill step, create/update task tracking for all steps and keep it synchronized as progress changes.
- User-question prompts mean to ask the user directly in Codex.
- Ignore Claude-specific mode-switch instructions when they appear.
- Strict execution contract: when a user explicitly invokes a skill, execute that skill protocol as written.
- Subagent authorization: when a skill is user-invoked or AI-detected and its protocol requires subagents, that skill activation authorizes use of the required
spawn_agent subagent(s) for that task.
- Do not skip, reorder, or merge protocol steps unless the user explicitly approves the deviation first.
- For workflow skills, execute each listed child-skill step explicitly and report step-by-step evidence.
- If a required step/tool cannot run in this environment, stop and ask the user before adapting.
Codex Project-Reference Loading (No Hooks)
Codex uses static project-reference loading instead of runtime-injected project docs.
When coding, planning, debugging, testing, or reviewing, open project docs explicitly using this routing.
Always read:
docs/project-config.json (project-specific paths, commands, modules, and workflow/test settings)
docs/project-reference/docs-index-reference.md (routes to the full docs/project-reference/* catalog)
docs/project-reference/lessons.md (always-on guardrails and anti-patterns)
Missing/stale context route: If docs/project-config.json, the docs index, lessons.md, CLAUDE.md, AGENTS.md, or any task-required reference doc is missing or stale, auto-run $project-init or the narrow setup route ($project-config, $docs-init, $scan-all, $scan --target=<key>, $claude-md-init) before ordinary project-specific work. If Codex mirrors or AGENTS.md are missing/stale, ask the user to run $sync-codex; do not auto-run it.
Situation-based docs:
- Project structure/architecture/tech-stack/deployment/setup (any layer — backend, frontend, or infra):
project-structure-reference.md
- Backend/CQRS/API/domain/entity changes:
backend-patterns-reference.md, domain-entities-reference.md
- Frontend/UI/styling/design-system:
frontend-patterns-reference.md, scss-styling-guide.md, design-system/README.md
- Spec authoring,
docs/specs/ pathing, or TC format: feature-spec-reference.md, spec-system-reference.md, spec-principles.md
- Behavior/public-contract changes or spec-test-code sync:
workflow-spec-test-code-cycle-reference.md plus the spec docs above
- Derived spec indexes/ERDs/reimplementation guides:
spec-system-reference.md and source Feature Specs under docs/specs/
- Integration test implementation/review:
integration-test-reference.md
- E2E test implementation/review:
e2e-test-reference.md
- Code review/audit work:
code-review-rules.md plus domain docs above based on changed files
Do not read all docs blindly. Start from docs-index-reference.md, then open only relevant files for the task.
[BLOCKING] Execute skill steps in declared order. NEVER skip, reorder, or merge steps without explicit user approval.
[BLOCKING] Before each step or sub-skill call, update task tracking: set in_progress when step starts, set completed when step ends.
[BLOCKING] Every completed/skipped step MUST include brief evidence or explicit skip reason.
[BLOCKING] If Task tools are unavailable, create and maintain an equivalent step-by-step plan tracker with the same status transitions.
Quick Summary
Goal: Turn a vague product idea into a validated, tech-agnostic, module-anchored backlog artifact ready for $refine to convert into a PBI — preserving problem intent without leaking solution or stack choices.
Summary:
- Purpose: capture a raw idea as a structured, validated backlog artifact — preserve problem intent, leak no solution/stack, hand a clean narrative to
$refine.
- Main steps/tasks (run in order): (1) Gather info — problem/value/users/scope; (2) Generate artifact
IDEA-{YYMMDD}-{NNN}, status draft, from idea-template.md; (3) Capture problem statement/value/users; (4) Detect project module via Glob("docs/specs/*/README.md") (silent; prompt only if ambiguous); (5) Load feature context (8-12K token budget — entities, BR-/TC- patterns); (6) Save to canonical path; (6.5) Discovery Interview — 3-5 ask the user directly; (7) Validate — 2-3 ask the user directly; (8) Suggest next → $refine.
- Two interview gates are NON-NEGOTIABLE: Discovery Interview (Step 6.5, 3-5 ask the user directly items incl. the always-on testability question) AND Validation (Step 7, 2-3 items) — never skip either even for "simple" ideas.
- Keep the problem statement strictly tech-agnostic (M1): name no framework/product/language/pattern, and do NOT assign logical IDs — the downstream PBI inherits the clean narrative and owns
FR-/BR- assignment.
- Auto-detect the project module silently via
Glob("docs/specs/*/README.md") and load feature context (8-12K token budget); prompt only when ambiguous. Greenfield (no real code dirs) → skip module detection and NEVER ask about tech stack.
- Persist to
team-artifacts/ideas/{YYMMDD}-{role}-idea-{slug}.md with t_shirt_size, then hand off to $refine for PBI conversion.
MANDATORY IMPORTANT MUST ATTENTION task tracking task to READ project-specific reference doc:
project-structure-reference.md — project patterns and structure. Not found → search: project documentation, coding standards, architecture docs.
Workflow:
- Gather Info — Ask problem, value, scope, target users
- Generate Artifact — Create idea file with ID (
IDEA-YYMMDD-NNN) + draft status
- Detect Module — Auto-match project module, load feature context from docs
- Discovery Interview — ask the user directly 3-5 structured questions (MANDATORY)
- Validate — Confirm problem statement, scope, stakeholders (MANDATORY)
- Suggest Next — Point to
$refine for PBI creation
Key Rules:
- Output:
team-artifacts/ideas/{YYMMDD}-{role}-idea-{slug}.md
- Validation NEVER optional — MANDATORY step
- Auto-detect module silently; prompt only when ambiguous
- MUST ATTENTION include
t_shirt_size (XS/S/M/L/XL) in artifact for early sizing
- [BLOCKING] Tech-agnostic output (M1): the problem statement stays tech-agnostic per
docs/project-reference/spec-principles.md §3 (all modes, not only greenfield) — name no framework/product/language/design-pattern; defer any stack preference to the later tech-research phase.
- M3 Logical-ID Assignment (forward to PBI): See
.claude/skills/shared/sdd-artifact-contract.md → "AI-SDD Mandates (M1-M7)" for BLOCKING criteria. An idea is a tech-agnostic business-intent definition only — do NOT assign logical IDs yet. When the idea advances to a PBI (via $refine), the PBI assigns logical IDs (FR-/BR-) as the PRIMARY citation spine and tracks [Source: namespace/service/id] abstract-anchor evidence (never physical code coordinates or repository-root paths) in a SEPARATE carrier from the business-intent prose. Keep the idea's problem/value narrative free of source identifiers so the PBI can inherit it cleanly.
Greenfield Mode
Auto-detected: No codebase found (no discovered source directories, no manifest files, no populated project-config.json) → greenfield mode. Planning artifacts (docs/, plans/, .claude/) don't count — repository must have actual code directories with content.
When greenfield detected:
- Skip module auto-detection (no modules exist yet)
- Skip
project-structure-reference.md read (won't exist)
- Focus broader problem-space: market gap, competitors, differentiation
- Output tech-agnostic problem statement
- Enable WebSearch for market/competitor context
- Increase ask the user directly frequency — capture vision, constraints, team profile, scale expectations
- [CRITICAL] NEVER ask about tech stack during idea capture. Tech stack = research-driven decision AFTER full business analysis (business-evaluation phase). User volunteers preference → acknowledge but defer to tech stack research phase.
Detailed Workflow
Step 1: Gather Information
- No title → ask: "What's the idea in one sentence?"
- Ask: "What problem does this solve?"
- Ask: "Who benefits from this?"
- Ask: "Any initial scope thoughts?"
Step 2: Generate Artifact
- Template:
.claude/docs/team-artifacts/templates/idea-template.md
- ID:
IDEA-{YYMMDD}-{NNN} (sequential)
- Status:
draft
Step 3: Capture Details
- Document problem statement, expected value, target users
Step 4: Detect Project Module
Dynamic Discovery:
- Run:
Glob("docs/specs/*/README.md")
- Extract module names from paths
- Match idea keywords against module keywords
| Scenario |
Action |
| Clear match |
Auto-detect — NEVER show confidence levels |
| Ambiguous / no match |
Prompt: "Which project module?" + Glob results + "Cross-cutting/Infrastructure" |
| 2+ modules detected |
Load ALL modules, add all to related_features |
If module detected:
- Read
docs/specs/{module}/README.md (first 200 lines)
- Extract feature list from Quick Navigation section
- Add to frontmatter:
module: {detected_module}, related_features: [Feature1, Feature2]
Step 5: Load Feature Context
- Read module README overview (~2K tokens)
- Identify closest matching feature(s)
- Read corresponding feature doc (3-5K tokens)
- Extract: related entities, existing business rules (BR-{MOD}-XXX), test case patterns (TC-{FEATURE}-{NNN})
Token Budget: Target 8-12K tokens total.
Step 6: Save Artifact
- Path:
team-artifacts/ideas/{YYMMDD}-{role}-idea-{slug}.md
- Role: infer from context or ask
- Include domain context if detected
Artifact Path (canonical convention) — Command $idea → base path team-artifacts/ideas/, role token po, type idea. Filename pattern: {YYMMDD}-{role}-{type}-{slug}.md → e.g. 260119-po-idea-dark-mode-toggle.md. Slug = lowercased basename, non-alphanumeric → -, trimmed, max 50 chars.
Step 6.5: Discovery Interview (MANDATORY)
Use ask the user directly — 3-5 structured questions. Each MUST ATTENTION have 2-4 options with one marked "(Recommended)".
| Category |
Purpose |
Example |
| Problem Clarity |
Distinguish problem from solution |
"What problem does this solve?" + options |
| User Persona |
Identify primary user |
"Who benefits most?" + role options |
| Scope |
MVP vs full vision |
"What's smallest valuable version?" |
| Testability |
Define done? |
"How would you verify this works?" |
| Impact |
Business value sizing |
"How many users/processes affected?" |
| Constraints |
Known blockers |
"Any technical/business constraints?" |
| Scale |
Expected load/growth |
"How many users/transactions expected?" |
Greenfield: NEVER include tech stack questions. Focus on business problem, users, scale, constraints.
Testability Question (ALWAYS include):
"How would you verify this feature works correctly?" — Options: manual test steps, automated test criteria, metric thresholds.
Document all answers under ## Discovery Interview.
Step 7: Validate Idea (MANDATORY)
ask the user directly — 2-3 validation questions:
| Category |
Example Question |
| Problem |
"Is the problem statement clear and user-focused?" |
| Value |
"What's the expected business value or user benefit?" |
| Scope |
"Any scope boundaries to clarify now?" |
| Stakeholders |
"Who else should review this idea?" |
Document under ## Validation Summary. Update artifact based on answers.
Validation Output Format:
## Validation Summary
**Validated:** {date}
### Confirmed
- {decision}: {user choice}
### Action Items
- [ ] {follow-up if any}
Step 8: Suggest Next Step
ask the user directly after capture:
$refine — Refine into PBI (Recommended)
$spec [mode=tests] — Jump straight to test spec
$plan — Start implementation planning
Output: "Idea captured! To refine into a PBI, run: $refine {filename}"
Module detected: "Module context from {module} will be used during refinement."
Output Formats
Domain Context Section
## Domain Context (Project Features)
### Module
{module_name}
### Related Features
- {Feature1} - [docs link]
- {Feature2} - [docs link]
### Domain Entities
- **Primary:** {Entity1}, {Entity2}
- **Related:** {Entity3}
### Existing Business Rules
- BR-{MOD}-XXX: {Brief description}
UI Sketch Section
## UI Sketch
### Layout
{Rough ASCII wireframe — see UI wireframe protocol}
### Key Components
- **{Component}** — {purpose} _(tier: common | domain-shared | page/app)_
Search existing libs before proposing new components.
Backend-only idea: ## UI Sketch → N/A — Backend-only change. No UI affected.
Examples
$idea "Dark mode toggle for settings"
# Creates: team-artifacts/ideas/260119-po-idea-dark-mode-toggle.md
$idea "Add goal progress tracking notification"
# Creates with module context: team-artifacts/ideas/260119-po-idea-goal-progress-notification.md
Workflow Recommendation
MANDATORY IMPORTANT MUST ATTENTION — NO EXCEPTIONS: Not already in workflow → MUST ATTENTION use ask the user directly:
- Activate
workflow-idea-to-pbi workflow (Recommended) — idea → refine → artifact-review --type=pbi → story → artifact-review --type=story → prioritize
- Execute
$idea directly — run standalone
Next Steps
MANDATORY IMPORTANT MUST ATTENTION — NO EXCEPTIONS after completing skill, use ask the user directly:
- "$refine (Recommended)" — Transform idea into actionable PBI
- "$web-research" — Idea needs market research first
- "Skip, continue manually" — user decides
[IMPORTANT] task tracking break ALL work into small tasks BEFORE starting — including tasks each file read. Simple tasks: AI MUST ATTENTION ask user whether to skip.
External Memory: Complex/lengthy work (research, analysis, scan, review) → write intermediate findings to plans/reports/ — prevents context loss, serves as deliverable.
Evidence Gate: MANDATORY IMPORTANT MUST ATTENTION — every claim, finding, recommendation requires file:line proof or traced evidence, confidence percentage (>80% act, <80% verify first).
UI Wireframe — Process visual design input (Figma URLs, screenshots, wireframes) via appropriate tool BEFORE creating wireframes. Use box-drawing ASCII characters for spatial layout. Classify every component into exactly ONE tier: Common (cross-app reusable) / Domain-Shared (cross-domain) / Page (single-page). Duplicate UI code = wrong tier. Search existing component libraries before creating new (>=80% match = reuse). Detail level varies by skill (idea=rough, story=full decomposition).
AI Mistake Prevention — Failure modes to avoid on every task:
Re-read files after context changes. Context compaction, resume, or long-running work can make memory stale; verify current files before acting.
Verify generated content against source evidence. AI hallucinates APIs, names, claims, and document facts. Check the relevant source before documenting or referencing.
Check downstream references before deleting or renaming. Removing an artifact can stale docs, generated mirrors, configs, and callers; map references first.
Trace the full impact chain after edits. Changing a definition can miss derived outputs and consumers. Follow the affected chain before declaring done.
Verify ALL affected outputs, not just the first. One green check is not all green checks; validate every output surface the change can affect.
Assume existing values are intentional — ask WHY before changing OR flagging one as a defect. Before changing or reporting a constant, limit, flag, cutoff, wording, or pattern, read nearby context and history, the CALLER's ordering, and 2+ sibling call sites of the same convention. A doc stating WHAT without WHY is missing rationale, not proof of a missing guard.
Surface ambiguity before acting — don't pick silently. Multiple valid interpretations require an explicit question or stated assumption with risk.
Assert the outcome your system owns, not the intermediate state your infrastructure owns. When verifying async work, assert the final business state — never the delivery/retry bookkeeping held in shared infrastructure that any co-running process can write. Such a check passes when run alone and flakes the moment anything else shares that infrastructure.
Keep shared guidance role-relevant. Universal guidance must help every receiving skill or agent; code-specific obligations belong only in code-specific protocols.
Critical Thinking Mindset — Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence >80% to act.
Anti-hallucination: Never present guess as fact — cite sources for every claim, admit uncertainty freely, self-check output for errors, cross-reference independently, stay skeptical of own confidence — certainty without evidence root of all hallucination.
Sequential Thinking Protocol — Structured multi-step reasoning for complex/ambiguous work. Use when planning, reviewing, debugging, or refining ideas where one-shot reasoning is unsafe.
Trigger when: complex problem decomposition · adaptive plans needing revision · analysis with course correction · unclear/emerging scope · multi-step solutions · hypothesis-driven debugging · cross-cutting trade-off evaluation.
Format (explicit mode — visible thought trail):
Thought N/M: [aspect] — one aspect per thought, state assumptions/uncertainty
Thought N/M [REVISION of Thought K]: ... — when prior reasoning invalidated; state Original / Why revised / Impact
Thought N/M [BRANCH A from Thought K]: ... — explore alternative; converge with decision rationale
Thought N/M [HYPOTHESIS]: ... then [VERIFICATION]: ... — test before acting
Thought N/N [FINAL] — only when verified, all critical aspects addressed, confidence >80%
Mandatory closers: Confidence % stated · Assumptions listed · Open questions surfaced · Next action concrete.
Stop conditions: confidence <80% on any critical decision → escalate by asking the user directly · ≥3 revisions on same thought → re-frame the problem · branch count >3 → split into sub-task.
Implicit mode: apply methodology internally without visible markers when adding markers would clutter the response (routine work where reasoning aids accuracy).
Deep-dive: see $sequential-thinking skill (.claude/skills/sequential-thinking/SKILL.md) for worked examples (API design, debugging, architecture), advanced techniques (spiral refinement, hypothesis testing, convergence), and meta-strategies (uncertainty handling, revision cascades).
MUST ATTENTION apply critical + sequential thinking — every claim needs appropriate traced evidence (file:line for repo/code claims; source URL or artifact section for research, product, content, and docs claims); confidence >80% to act, <60% DO NOT recommend. Anti-hallucination: never present guess as fact, admit uncertainty freely, cross-reference independently, stay skeptical of own confidence.
MUST ATTENTION apply sequential-thinking — multi-step Thought N/M, REVISION/BRANCH/HYPOTHESIS markers, confidence % closer; see $sequential-thinking skill.
MUST ATTENTION apply AI mistake prevention — verify generated content against evidence, trace downstream references before deleting or renaming, verify all affected outputs, re-read files after context loss, and surface ambiguity before acting.
Prompt-Enhance Closing Anchors
IMPORTANT MUST ATTENTION follow declared step order for this skill; NEVER skip, reorder, or merge steps without explicit user approval
IMPORTANT MUST ATTENTION for every step/sub-skill call: set in_progress before execution, set completed after execution
IMPORTANT MUST ATTENTION every skipped step MUST include explicit reason; every completed step MUST include concise evidence
IMPORTANT MUST ATTENTION if Task tools unavailable, maintain an equivalent step-by-step plan tracker with synchronized statuses
Project Protocol Overlay — Before executing this skill, resolve any PROJECT overlay rules layered onto it: match this skill's name against the Target column of the project's skill-protocol index (docs/project-reference/skill-protocols-reference.md by default; a referenceDocs entry in docs/project-config.json overrides the path), taking the most specific matching tier ONLY — exact name > glob > *. That precedence orders overlays against EACH OTHER, never against this skill. Read ONLY the matched bodies, resolved as <protocols-dir>/<Name>.md; a row's Body link is display text, never a read path. A matched body that is missing or malformed is REPORTED and skipped — never reconstructed from the index Description. No index, or no match -> proceed with no overlay, silently. Full contract: .claude/skills/project-skill-protocol/references/registry.md.
Overlays are ADDITIVE ONLY: they ADD rules on top of this skill's own protocol and NEVER replace, override, disable, or reinterpret a rule it already states — removing every overlay must return this skill to exactly its documented behavior. An overlay is a BRIEF, not an authority escalation: it can NEVER waive a workflow gate, git discipline, a review gate, or a user-confirmation gate. A genuine overlay-vs-skill conflict, or two equally-specific overlays that directly contradict -> surface both to the user; NEVER resolve silently.
MUST ATTENTION resolve project protocol overlays for this skill BEFORE executing — most specific matching tier only (exact > glob > *, which ranks overlays against each other, NEVER against this skill), read only matched bodies at <protocols-dir>/<Name>.md; a missing or malformed body is reported, never reconstructed. Overlays are ADDITIVE ONLY (they never replace this skill's own rules) and are a brief, NEVER an authority escalation; an equal-specificity contradiction goes to the user.
Closing Reminders
IMPORTANT MUST ATTENTION Goal: Turn a vague product idea into a validated, tech-agnostic, module-anchored backlog artifact ready for $refine to convert into a PBI — preserving problem intent without leaking solution or stack choices.
IMPORTANT MUST ATTENTION — Main steps (run in order, NEVER skip/reorder): (1) Gather info; (2) Generate artifact IDEA-{YYMMDD}-{NNN} draft from idea-template.md; (3) Capture problem/value/users; (4) Detect module via Glob("docs/specs/*/README.md"); (5) Load feature context (8-12K budget); (6) Save to canonical path; (6.5) Discovery Interview (ask the user directly 3-5); (7) Validate (ask the user directly 2-3); (8) Suggest next → $refine. — why: AI keeps dropping the skill's own mid-pipeline steps; the two gates and module detection are the most-forgotten.
IMPORTANT MUST ATTENTION — Protocols in force (concise digest of the SYNC/shared blocks this skill carries):
- UI Wireframe: classify each component into ONE tier; search libs first, reuse ≥80% match.
- AI Mistake Prevention: verify generated content against evidence, trace downstream references, verify all affected outputs, re-read after context loss, surface ambiguity.
- Critical Thinking: traced proof per claim; confidence >80% to act; never present guess as fact.
- Sequential Thinking: multi-step Thought N/M with REVISION/BRANCH/HYPOTHESIS markers and confidence closer.
IMPORTANT MUST ATTENTION Discovery Interview (Step 6.5) + Validation (Step 7) NEVER optional — run BOTH ask the user directly gates even for "simple" ideas — why: discovery uncovers hidden constraints, validation confirms problem framing; different question categories
IMPORTANT MUST ATTENTION ALWAYS keep problem statement tech-agnostic (M1, spec-principles.md §3, all modes) — name no framework/product/language/design-pattern; defer any stack preference to the later tech-research phase — why: PBI inherits the narrative cleanly downstream
IMPORTANT MUST ATTENTION in greenfield mode NEVER ask about tech stack — acknowledge a volunteered preference, then defer to the business-evaluation phase — why: stack is a research-driven decision after business analysis, not a capture-time guess
IMPORTANT MUST ATTENTION task tracking break ALL work into small tasks BEFORE starting — including a task to READ project-structure-reference.md (skip in greenfield — it won't exist)
IMPORTANT MUST ATTENTION validate all decisions with user by asking the user directly — NEVER auto-decide — and NEVER show confidence levels on an auto-detected module match
IMPORTANT MUST ATTENTION auto-detect module silently via Glob("docs/specs/*/README.md") — prompt only when ambiguous or no match; greenfield → skip module detection — why: confirm with Glob() evidence, not assumption
IMPORTANT MUST ATTENTION assign NO logical IDs (M3) — an idea is tech-agnostic business intent only; the downstream PBI owns FR-/BR- assignment and [Source: namespace/service/id] anchors — why: keep the problem/value narrative free of source identifiers so the PBI inherits it cleanly
IMPORTANT MUST ATTENTION include t_shirt_size (XS/S/M/L/XL) in the artifact and keep the feature-context load within the 8-12K token budget — why: early sizing feeds prioritization; over-budget reads dilute attention
IMPORTANT MUST ATTENTION persist to team-artifacts/ideas/{YYMMDD}-{role}-idea-{slug}.md, then hand off to $refine for PBI conversion — why: canonical path keeps downstream tooling aligned
IMPORTANT MUST ATTENTION search existing component libraries before proposing any new UI component (≥80% match = reuse); classify each into exactly ONE tier — why: duplicate UI code = wrong tier
IMPORTANT MUST ATTENTION cite file:line proof or traced evidence for every claim/recommendation, confidence >80% to act, <80% verify first — why: certainty without evidence is the root of hallucination
IMPORTANT MUST ATTENTION add a final review task to verify work quality
Anti-Rationalization:
| Evasion |
Rebuttal |
| "Idea is simple, skip interview" |
NEVER skip — discovery uncovers hidden constraints |
| "Module is obvious, skip detection" |
Still run Glob() — confirm with evidence not assumption |
| "Validation is redundant after interview" |
ALWAYS run both — different question categories |
| "Greenfield check is optional" |
Auto-detect is MANDATORY — no manual override |
| "User mentioned a framework, capture it" |
Stay tech-agnostic — acknowledge, defer to tech-research phase |
| "I'll assign FR-/BR- IDs now" |
NO logical IDs at idea stage — the PBI assigns them downstream |
| "Reuse an existing component? new is faster" |
Search libs first — ≥80% match = reuse; new without search = wrong tier |
[TASK-PLANNING] Before acting, analyze task scope and systematically break it into small todo tasks and sub-tasks using task tracking.
IMPORTANT MUST ATTENTION the 3 rules to never skip: (1) run BOTH Discovery + Validation ask the user directly gates; (2) keep the problem statement tech-agnostic (no stack/IDs); (3) cite file:line evidence, confidence >80% to act.
Hookless Prompt Protocol Mirror (Auto-Synced)
Source: .claude/.ck.json + .claude/skills/shared/sync-inline-versions.md (:full blocks) + .claude/scripts/lib/hookless-prompt-protocol.cjs
[WORKFLOW-EXECUTION-PROTOCOL] [BLOCKING] Workflow Execution Protocol — MANDATORY IMPORTANT MUST CRITICAL. Do not skip for any reason.
Generic portability boundary: Reusable skills and protocol text stay project-neutral; project-specific conventions are discovered from docs/project-config.json and docs/project-reference/. Apply shared AI-SDD from shared/sdd-artifact-contract.md. Read docs/project-config.json and docs/project-reference/docs-index-reference.md, then open the project reference docs named there. For spec, test-case, behavior-change, public-contract, or docs/specs/ work, route through the local spec docs named by the docs index: feature-spec-reference.md, spec-system-reference.md, spec-principles.md, and workflow-spec-test-code-cycle-reference.md when specs/tests/code must stay synchronized. If either file or a required reference doc is missing or stale, auto-run $project-init (or the narrow lower-level route such as $project-config, $docs-init, $scan-all, or $scan --target=<key>) before ordinary project-specific work. Any supported AI tool may execute when this shared context and local docs are available.
- DETECT: If the prompt starts with an explicit slash skill/workflow command, execute it directly. Otherwise match the prompt against the workflow catalog and skill list.
- ANALYZE: Choose the best option: execute directly, invoke a skill, activate a standard workflow, or compose a custom step combination.
- AUTO-SELECT: Pick the best option yourself. Do not ask the user to choose between direct execution, skill, standard workflow, or custom workflow.
- ACTIVATE: For a selected workflow, call
$start-workflow <workflowId>; for a selected skill, invoke that skill; for a custom workflow, sequence custom steps directly; for direct execution, proceed with the task.
- CREATE TASKS: task tracking for ALL workflow/skill/custom steps before execution when the selected path has multiple steps.
- PARALLELIZE: Before executing the task list, tag each task
PAR (independent inputs + write set disjoint from every other PAR task) or SEQ (name the blocking dependency), group PAR tasks into waves, declare the wave plan, and spawn each wave's sub-agents in ONE message — all-return barrier per wave, fan-out one level deep unless a sub-agent's own definition authorizes further fan-out. Sequential-by-default is a defect when tasks are independent; do not parallelize shared write targets, output-consuming tasks, trivial single-file work, ordering a skill or workflow explicitly fixes, or user-approval gates.
- EXECUTE: Advance per the Workflow Step Advancement & Parallel Phases rule in your context instructions — model-driven; a sub-agent completion advances a step identically to an inline call; a parallel-phase group is an all-return barrier (advance only after ALL members return, never serialize it)
Shared AI-SDD Protocol Markers
Source: .claude/skills/shared/sync-inline-versions.md
SYNC:ai-sdd-artifact-contract
AI-SDD Artifact Contract — Shared spec-driven development rules stay portable and source-owned.
- Keep reusable AI-SDD principles in
.claude; put repository-specific paths, commands, owners, products, and formats in project config/reference docs.
- Preserve cycle:
spec -> plan -> tasks -> implement -> verify -> update spec/docs.
- Trace every requirement or invariant through decision, task, TC/test, source evidence, and docs/spec update.
- Treat code-to-spec extraction as reference-only until accepted by the canonical spec owner.
- Any supported AI tool may plan, implement, review, or verify with synced context; using multiple tools is optional.
- Update
.claude source first, then sync generated mirrors; do not manually edit .agents, .codex, or AGENTS.md. — why: mirrors are generated artifacts; hand-edits are overwritten on the next sync
- If
docs/project-config.json, root instruction files, or a required project-reference doc is missing or stale, auto-run $project-init or the narrow lower-level route before ordinary project-specific work.
Active reference: shared/sdd-artifact-contract.md in the active skills root.
SYNC:ai-sdd-artifact-contract:reminder
- MANDATORY Apply
shared/sdd-artifact-contract.md; keep reusable AI-SDD in .claude and local rules in project docs.
- MANDATORY Code-to-spec extraction is reference-only until canonical acceptance; any supported AI tool may execute with synced context.
- MANDATORY Update
.claude source before syncing generated mirrors; do not manually edit .agents, .codex, or AGENTS.md.
- MANDATORY Missing or stale project config, root instruction files, or required reference docs route project-specific work through
$project-init or the narrow setup route automatically.
[TASK-PLANNING] [MANDATORY] BEFORE executing any workflow or skill step, create/update task tracking for all planned steps, then keep it synchronized as each step starts/completes.
[LESSON-LEARNED-REMINDER] [BLOCKING] Task Planning & Continuous Improvement — MANDATORY. Do not skip.
Break work into small tasks (task tracking) before starting. Add final task: "Analyze AI mistakes & lessons learned".
Extract lessons — ROOT CAUSE ONLY, not symptom fixes:
- Name the FAILURE MODE (reasoning/assumption failure), not symptom — "assumed API existed without reading source" not "used wrong enum value".
- Generality test: does this failure mode apply to ≥3 contexts/codebases? If not, abstract one level up.
- Write as a universal rule — strip project-specific names/paths/classes. Useful on any codebase.
- Consolidate: multiple mistakes sharing one failure mode → ONE lesson.
- Recurrence gate: "Would this recur in future session WITHOUT this reminder?" — No → skip
$learn.
- Auto-fix gate: "Could
$code-review/$code-simplifier/$security-review/$lint catch this?" — Yes → improve review skill instead.
- BOTH gates pass → ask user to run
$learn.
[CRITICAL-THINKING-MINDSET] Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence >80% to act.
Anti-hallucination principle: Never present guess as fact — cite sources for every claim, admit uncertainty freely, self-check output for errors, cross-reference independently, stay skeptical of own confidence — certainty without evidence root of all hallucination.
AI Attention principle (Primacy-Recency): Put the 3 most critical rules at both top and bottom of long prompts/protocols so instruction adherence survives long context windows.
Goal-driven execution: Define success criteria first, loop until verified, and stop only when observable checks pass.
Tests verify intent: Tests must protect business rules/invariants and fail when the protected intent breaks, not only mirror current behavior.
Common AI Mistake Prevention (System Lessons)
- Re-read files after context compaction. Edit requires prior Read in same context; compaction wipes read state. Re-read before editing.
- Grep for old terms after bulk replacements. AI over-trusts find/replace completeness. Grep full repo after bulk edits for missed refs in docs/configs/catalogs.
- Check downstream references before deleting. Deletions cascade doc/code staleness. Map referencing files before removal.
- After memory loss, check existing state before creating new. Compaction wipes prior-work memory. Query current state to resume — never blindly duplicate.
- Verify AI-generated content against actual code. AI hallucinates APIs, class names, method signatures. Grep to confirm existence before documenting/referencing.
- Trace full dependency chain after edits. Changing a definition misses downstream consumers. Trace the full chain.
- When renaming, grep ALL consumer file types. Some file types silently ignore missing refs (no compile error). Search code, templates, configs, generated files.
- Trace ALL code paths when verifying correctness. Code existing ≠ code executing. Trace early exits, error branches, conditional skips — not just happy path.
- Update docs that embed canonical data when source changes. Docs inlining derived data (workflows, schemas, configs) go stale silently. Update all embedding docs alongside source.
- Verify sub-agent results after context recovery. Background agents may finish while parent compacted — grep-verify output, don't trust assumed completion.
- Cross-check full target list against sub-agent assignments. Parallel sub-agents by category miss boundary items. Reconcile union of assignments against target list before proceeding.
- Sub-agents inherit knowledge only from their agent .md definition — use custom agent types, not built-in Explore. Tool adoption = permission + knowledge + enforcement (numbered workflow step).
- Persist sub-agent findings incrementally, not as a final batch. Long sub-agents hit cutoffs before final write — findings lost. Instruct append-per-section to report file.
- When debugging, ask "whose responsibility?" before fixing. Trace caller (wrong data) vs callee (wrong handling). Fix at responsible layer — never patch symptom site.
- Test failure → record a provisional verdict before trace/edit, then investigate. Use the full five-way taxonomy: SOURCE-WRONG (production violates intent), TEST-WRONG (assertion/setup is stale), TEST-NOT-OPTIMAL (valid but fragile or low-signal test), ENVIRONMENT-BLOCKED (external state prevents a verdict), or AMBIGUOUS (intent/evidence cannot choose safely). Then trace root cause and triangulate against the governing spec (
docs/specs/** if one exists) AND source. NEVER weaken an assertion, add a skip, relax a timeout, or change source merely to force green.
- **Grep ALL removed names after extraction
…(truncated)
1---
2name: idea
3description: '[Project Management] Use when capturing new ideas, feature requests, or concepts for future refinement.'
4metadata:
5 source: https://skillmd.com/skills/duc01226/idea
6---
7
8> Codex compatibility note:
9>
10> - Invoke repository skills with `$skill-name` in Codex; this mirrored copy rewrites legacy Claude `/skill-name` references.
11> - Task tracker mandate: BEFORE executing any workflow or skill step, create/update task tracking for all steps and keep it synchronized as progress changes.
12> - User-question prompts mean to ask the user directly in Codex.
13> - Ignore Claude-specific mode-switch instructions when they appear.
14> - Strict execution contract: when a user explicitly invokes a skill, execute that skill protocol as written.
15> - Subagent authorization: when a skill is user-invoked or AI-detected and its protocol requires subagents, that skill activation authorizes use of the required `spawn_agent` subagent(s) for that task.
16> - Do not skip, reorder, or merge protocol steps unless the user explicitly approves the deviation first.
17> - For workflow skills, execute each listed child-skill step explicitly and report step-by-step evidence.
18> - If a required step/tool cannot run in this environment, stop and ask the user before adapting.
19
20<!-- CODEX:PROJECT-REFERENCE-LOADING:START -->
21
22## Codex Project-Reference Loading (No Hooks)
23
24Codex uses static project-reference loading instead of runtime-injected project docs.
25When coding, planning, debugging, testing, or reviewing, open project docs explicitly using this routing.
26
27**Always read:**
28
29- `docs/project-config.json` (project-specific paths, commands, modules, and workflow/test settings)
30- `docs/project-reference/docs-index-reference.md` (routes to the full `docs/project-reference/*` catalog)
31- `docs/project-reference/lessons.md` (always-on guardrails and anti-patterns)
32
33**Missing/stale context route:** If `docs/project-config.json`, the docs index, `lessons.md`, `CLAUDE.md`, `AGENTS.md`, or any task-required reference doc is missing or stale, auto-run `$project-init` or the narrow setup route (`$project-config`, `$docs-init`, `$scan-all`, `$scan --target=<key>`, `$claude-md-init`) before ordinary project-specific work. If Codex mirrors or `AGENTS.md` are missing/stale, ask the user to run `$sync-codex`; do not auto-run it.
34
35**Situation-based docs:**
36
37- Project structure/architecture/tech-stack/deployment/setup (any layer — backend, frontend, or infra): `project-structure-reference.md`
38- Backend/CQRS/API/domain/entity changes: `backend-patterns-reference.md`, `domain-entities-reference.md`
39- Frontend/UI/styling/design-system: `frontend-patterns-reference.md`, `scss-styling-guide.md`, `design-system/README.md`
40- Spec authoring, `docs/specs/` pathing, or TC format: `feature-spec-reference.md`, `spec-system-reference.md`, `spec-principles.md`
41- Behavior/public-contract changes or spec-test-code sync: `workflow-spec-test-code-cycle-reference.md` plus the spec docs above
42- Derived spec indexes/ERDs/reimplementation guides: `spec-system-reference.md` and source Feature Specs under `docs/specs/`
43- Integration test implementation/review: `integration-test-reference.md`
44- E2E test implementation/review: `e2e-test-reference.md`
45- Code review/audit work: `code-review-rules.md` plus domain docs above based on changed files
46
47Do not read all docs blindly. Start from `docs-index-reference.md`, then open only relevant files for the task.
48
49<!-- CODEX:PROJECT-REFERENCE-LOADING:END -->
50
51<!-- PROMPT-ENHANCE:STEP-TASK-ANCHOR:START -->
52
53> **[BLOCKING]** Execute skill steps in declared order. NEVER skip, reorder, or merge steps without explicit user approval.
54> **[BLOCKING]** Before each step or sub-skill call, update task tracking: set `in_progress` when step starts, set `completed` when step ends.
55> **[BLOCKING]** Every completed/skipped step MUST include brief evidence or explicit skip reason.
56> **[BLOCKING]** If Task tools are unavailable, create and maintain an equivalent step-by-step plan tracker with the same status transitions.
57
58<!-- PROMPT-ENHANCE:STEP-TASK-ANCHOR:END -->
59
60## Quick Summary
61
62**Goal:** Turn a vague product idea into a validated, tech-agnostic, module-anchored backlog artifact ready for `$refine` to convert into a PBI — preserving problem intent without leaking solution or stack choices.
63
64**Summary:**
65
66- **Purpose:** capture a raw idea as a structured, validated backlog artifact — preserve problem intent, leak no solution/stack, hand a clean narrative to `$refine`.
67- **Main steps/tasks (run in order):** (1) Gather info — problem/value/users/scope; (2) Generate artifact `IDEA-{YYMMDD}-{NNN}`, status `draft`, from `idea-template.md`; (3) Capture problem statement/value/users; (4) Detect project module via `Glob("docs/specs/*/README.md")` (silent; prompt only if ambiguous); (5) Load feature context (8-12K token budget — entities, BR-/TC- patterns); (6) Save to canonical path; (6.5) **Discovery Interview** — 3-5 ask the user directly; (7) **Validate** — 2-3 ask the user directly; (8) Suggest next → `$refine`.
68- Two interview gates are NON-NEGOTIABLE: Discovery Interview (Step 6.5, 3-5 ask the user directly items incl. the always-on testability question) AND Validation (Step 7, 2-3 items) — never skip either even for "simple" ideas.
69- Keep the problem statement strictly tech-agnostic (M1): name no framework/product/language/pattern, and do NOT assign logical IDs — the downstream PBI inherits the clean narrative and owns `FR-`/`BR-` assignment.
70- Auto-detect the project module silently via `Glob("docs/specs/*/README.md")` and load feature context (8-12K token budget); prompt only when ambiguous. Greenfield (no real code dirs) → skip module detection and NEVER ask about tech stack.
71- Persist to `team-artifacts/ideas/{YYMMDD}-{role}-idea-{slug}.md` with `t_shirt_size`, then hand off to `$refine` for PBI conversion.
72
73> **MANDATORY IMPORTANT MUST ATTENTION** task tracking task to READ project-specific reference doc:
74> `project-structure-reference.md` — project patterns and structure. Not found → search: project documentation, coding standards, architecture docs.
75
76**Workflow:**
77
781. **Gather Info** — Ask problem, value, scope, target users
792. **Generate Artifact** — Create idea file with ID (`IDEA-YYMMDD-NNN`) + draft status
803. **Detect Module** — Auto-match project module, load feature context from docs
814. **Discovery Interview** — ask the user directly 3-5 structured questions (MANDATORY)
825. **Validate** — Confirm problem statement, scope, stakeholders (MANDATORY)
836. **Suggest Next** — Point to `$refine` for PBI creation
84
85**Key Rules:**
86
87- Output: `team-artifacts/ideas/{YYMMDD}-{role}-idea-{slug}.md`
88- Validation NEVER optional — MANDATORY step
89- Auto-detect module silently; prompt only when ambiguous
90- MUST ATTENTION include `t_shirt_size` (XS/S/M/L/XL) in artifact for early sizing
91- **[BLOCKING] Tech-agnostic output (M1):** the problem statement stays tech-agnostic per `docs/project-reference/spec-principles.md` §3 (all modes, not only greenfield) — name no framework/product/language/design-pattern; defer any stack preference to the later tech-research phase.
92- **M3 Logical-ID Assignment (forward to PBI):** See `.claude/skills/shared/sdd-artifact-contract.md` → "AI-SDD Mandates (M1-M7)" for BLOCKING criteria. An idea is a tech-agnostic business-intent definition only — do NOT assign logical IDs yet. When the idea advances to a PBI (via `$refine`), the PBI assigns logical IDs (`FR-`/`BR-`) as the PRIMARY citation spine and tracks `[Source: namespace/service/id]` abstract-anchor evidence (never physical code coordinates or repository-root paths) in a SEPARATE carrier from the business-intent prose. Keep the idea's problem/value narrative free of source identifiers so the PBI can inherit it cleanly.
93
94## Greenfield Mode
95
96> **Auto-detected:** No codebase found (no discovered source directories, no manifest files, no populated `project-config.json`) → greenfield mode. Planning artifacts (`docs/`, `plans/`, `.claude/`) don't count — repository must have actual code directories with content.
97
98**When greenfield detected:**
99
1001. Skip module auto-detection (no modules exist yet)
1012. Skip `project-structure-reference.md` read (won't exist)
1023. Focus broader problem-space: market gap, competitors, differentiation
1034. Output tech-agnostic problem statement
1045. Enable WebSearch for market/competitor context
1056. Increase ask the user directly frequency — capture vision, constraints, team profile, scale expectations
1067. **[CRITICAL] NEVER ask about tech stack during idea capture.** Tech stack = research-driven decision AFTER full business analysis (business-evaluation phase). User volunteers preference → acknowledge but defer to tech stack research phase.
107
108## Detailed Workflow
109
110### Step 1: Gather Information
111
112- No title → ask: "What's the idea in one sentence?"
113- Ask: "What problem does this solve?"
114- Ask: "Who benefits from this?"
115- Ask: "Any initial scope thoughts?"
116
117### Step 2: Generate Artifact
118
119- Template: `.claude/docs/team-artifacts/templates/idea-template.md`
120- ID: `IDEA-{YYMMDD}-{NNN}` (sequential)
121- Status: `draft`
122
123### Step 3: Capture Details
124
125- Document problem statement, expected value, target users
126
127### Step 4: Detect Project Module
128
129**Dynamic Discovery:**
130
1311. Run: `Glob("docs/specs/*/README.md")`
1322. Extract module names from paths
1333. Match idea keywords against module keywords
134
135| Scenario | Action |
136| -------------------- | ------------------------------------------------------------------------------- |
137| Clear match | Auto-detect — NEVER show confidence levels |
138| Ambiguous / no match | Prompt: "Which project module?" + Glob results + "Cross-cutting/Infrastructure" |
139| 2+ modules detected | Load ALL modules, add all to `related_features` |
140
141**If module detected:**
142
1431. Read `docs/specs/{module}/README.md` (first 200 lines)
1442. Extract feature list from Quick Navigation section
1453. Add to frontmatter: `module: {detected_module}`, `related_features: [Feature1, Feature2]`
146
147### Step 5: Load Feature Context
148
1491. Read module README overview (~2K tokens)
1502. Identify closest matching feature(s)
1513. Read corresponding feature doc (3-5K tokens)
1524. Extract: related entities, existing business rules (BR-{MOD}-XXX), test case patterns (TC-{FEATURE}-{NNN})
153
154**Token Budget:** Target 8-12K tokens total.
155
156### Step 6: Save Artifact
157
158- Path: `team-artifacts/ideas/{YYMMDD}-{role}-idea-{slug}.md`
159- Role: infer from context or ask
160- Include domain context if detected
161
162> **Artifact Path (canonical convention)** — Command `$idea` → base path `team-artifacts/ideas/`, role token `po`, type `idea`. Filename pattern: `{YYMMDD}-{role}-{type}-{slug}.md` → e.g. `260119-po-idea-dark-mode-toggle.md`. Slug = lowercased basename, non-alphanumeric → `-`, trimmed, max 50 chars.
163
164### Step 6.5: Discovery Interview (MANDATORY)
165
166Use ask the user directly — 3-5 structured questions. Each MUST ATTENTION have 2-4 options with one marked "(Recommended)".
167
168| Category | Purpose | Example |
169| --------------- | --------------------------------- | ----------------------------------------- |
170| Problem Clarity | Distinguish problem from solution | "What problem does this solve?" + options |
171| User Persona | Identify primary user | "Who benefits most?" + role options |
172| Scope | MVP vs full vision | "What's smallest valuable version?" |
173| Testability | Define done? | "How would you verify this works?" |
174| Impact | Business value sizing | "How many users/processes affected?" |
175| Constraints | Known blockers | "Any technical/business constraints?" |
176| Scale | Expected load/growth | "How many users/transactions expected?" |
177
178> **Greenfield:** NEVER include tech stack questions. Focus on business problem, users, scale, constraints.
179
180**Testability Question (ALWAYS include):**
181"How would you verify this feature works correctly?" — Options: manual test steps, automated test criteria, metric thresholds.
182
183Document all answers under `## Discovery Interview`.
184
185### Step 7: Validate Idea (MANDATORY)
186
187ask the user directly — 2-3 validation questions:
188
189| Category | Example Question |
190| ------------ | ----------------------------------------------------- |
191| Problem | "Is the problem statement clear and user-focused?" |
192| Value | "What's the expected business value or user benefit?" |
193| Scope | "Any scope boundaries to clarify now?" |
194| Stakeholders | "Who else should review this idea?" |
195
196Document under `## Validation Summary`. Update artifact based on answers.
197
198**Validation Output Format:**
199
200```markdown
201## Validation Summary
202
203**Validated:** {date}
204
205### Confirmed
206
207- {decision}: {user choice}
208
209### Action Items
210
211- [ ] {follow-up if any}
212```
213
214### Step 8: Suggest Next Step
215
216ask the user directly after capture:
217
2181. `$refine` — Refine into PBI (Recommended)
2192. `$spec [mode=tests]` — Jump straight to test spec
2203. `$plan` — Start implementation planning
221
222Output: "Idea captured! To refine into a PBI, run: `$refine {filename}`"
223Module detected: "Module context from {module} will be used during refinement."
224
225## Output Formats
226
227### Domain Context Section
228
229```markdown
230## Domain Context (Project Features)
231
232### Module
233
234{module_name}
235
236### Related Features
237
238- {Feature1} - [docs link]
239- {Feature2} - [docs link]
240
241### Domain Entities
242
243- **Primary:** {Entity1}, {Entity2}
244- **Related:** {Entity3}
245
246### Existing Business Rules
247
248- BR-{MOD}-XXX: {Brief description}
249```
250
251### UI Sketch Section
252
253```markdown
254## UI Sketch
255
256### Layout
257
258{Rough ASCII wireframe — see UI wireframe protocol}
259
260### Key Components
261
262- **{Component}** — {purpose} _(tier: common | domain-shared | page/app)_
263```
264
265> Search existing libs before proposing new components.
266> Backend-only idea: `## UI Sketch` → `N/A — Backend-only change. No UI affected.`
267
268## Examples
269
270```bash
271$idea "Dark mode toggle for settings"
272# Creates: team-artifacts/ideas/260119-po-idea-dark-mode-toggle.md
273
274$idea "Add goal progress tracking notification"
275# Creates with module context: team-artifacts/ideas/260119-po-idea-goal-progress-notification.md
276```
277
278---
279
280## Workflow Recommendation
281
282> **MANDATORY IMPORTANT MUST ATTENTION — NO EXCEPTIONS:** Not already in workflow → MUST ATTENTION use ask the user directly:
283>
284> 1. **Activate `workflow-idea-to-pbi` workflow** (Recommended) — idea → refine → artifact-review --type=pbi → story → artifact-review --type=story → prioritize
285> 2. **Execute `$idea` directly** — run standalone
286
287---
288
289## Next Steps
290
291**MANDATORY IMPORTANT MUST ATTENTION — NO EXCEPTIONS** after completing skill, use ask the user directly:
292
293- **"$refine (Recommended)"** — Transform idea into actionable PBI
294- **"$web-research"** — Idea needs market research first
295- **"Skip, continue manually"** — user decides
296
297---
298
299> **[IMPORTANT]** task tracking break ALL work into small tasks BEFORE starting — including tasks each file read. Simple tasks: AI MUST ATTENTION ask user whether to skip.
300
301> **External Memory:** Complex/lengthy work (research, analysis, scan, review) → write intermediate findings to `plans/reports/` — prevents context loss, serves as deliverable.
302
303> **Evidence Gate:** MANDATORY IMPORTANT MUST ATTENTION — every claim, finding, recommendation requires `file:line` proof or traced evidence, confidence percentage (>80% act, <80% verify first).
304
305<!-- SYNC:ui-wireframe -->
306
307> **UI Wireframe** — Process visual design input (Figma URLs, screenshots, wireframes) via appropriate tool BEFORE creating wireframes. Use box-drawing ASCII characters for spatial layout. Classify every component into exactly ONE tier: Common (cross-app reusable) / Domain-Shared (cross-domain) / Page (single-page). Duplicate UI code = wrong tier. Search existing component libraries before creating new (>=80% match = reuse). Detail level varies by skill (idea=rough, story=full decomposition).
308
309<!-- /SYNC:ui-wireframe -->
310
311<!-- SYNC:ai-mistake-prevention -->
312
313> **AI Mistake Prevention** — Failure modes to avoid on every task:
314>
315> **Re-read files after context changes.** Context compaction, resume, or long-running work can make memory stale; verify current files before acting.
316> **Verify generated content against source evidence.** AI hallucinates APIs, names, claims, and document facts. Check the relevant source before documenting or referencing.
317> **Check downstream references before deleting or renaming.** Removing an artifact can stale docs, generated mirrors, configs, and callers; map references first.
318> **Trace the full impact chain after edits.** Changing a definition can miss derived outputs and consumers. Follow the affected chain before declaring done.
319> **Verify ALL affected outputs, not just the first.** One green check is not all green checks; validate every output surface the change can affect.
320> **Assume existing values are intentional — ask WHY before changing OR flagging one as a defect.** Before changing or reporting a constant, limit, flag, cutoff, wording, or pattern, read nearby context and history, the CALLER's ordering, and 2+ sibling call sites of the same convention. A doc stating WHAT without WHY is missing rationale, not proof of a missing guard.
321> **Surface ambiguity before acting — don't pick silently.** Multiple valid interpretations require an explicit question or stated assumption with risk.
322> **Assert the outcome your system owns, not the intermediate state your infrastructure owns.** When verifying async work, assert the final business state — never the delivery/retry bookkeeping held in shared infrastructure that any co-running process can write. Such a check passes when run alone and flakes the moment anything else shares that infrastructure.
323> **Keep shared guidance role-relevant.** Universal guidance must help every receiving skill or agent; code-specific obligations belong only in code-specific protocols.
324
325<!-- /SYNC:ai-mistake-prevention -->
326
327<!-- SYNC:critical-thinking-mindset -->
328
329> **Critical Thinking Mindset** — Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence >80% to act.
330> **Anti-hallucination:** Never present guess as fact — cite sources for every claim, admit uncertainty freely, self-check output for errors, cross-reference independently, stay skeptical of own confidence — certainty without evidence root of all hallucination.
331
332<!-- /SYNC:critical-thinking-mindset -->
333
334<!-- SYNC:sequential-thinking-protocol -->
335
336> **Sequential Thinking Protocol** — Structured multi-step reasoning for complex/ambiguous work. Use when planning, reviewing, debugging, or refining ideas where one-shot reasoning is unsafe.
337>
338> **Trigger when:** complex problem decomposition · adaptive plans needing revision · analysis with course correction · unclear/emerging scope · multi-step solutions · hypothesis-driven debugging · cross-cutting trade-off evaluation.
339>
340> **Format (explicit mode — visible thought trail):**
341>
342> 1. `Thought N/M: [aspect]` — one aspect per thought, state assumptions/uncertainty
343> 2. `Thought N/M [REVISION of Thought K]: ...` — when prior reasoning invalidated; state Original / Why revised / Impact
344> 3. `Thought N/M [BRANCH A from Thought K]: ...` — explore alternative; converge with decision rationale
345> 4. `Thought N/M [HYPOTHESIS]: ...` then `[VERIFICATION]: ...` — test before acting
346> 5. `Thought N/N [FINAL]` — only when verified, all critical aspects addressed, confidence >80%
347>
348> **Mandatory closers:** Confidence % stated · Assumptions listed · Open questions surfaced · Next action concrete.
349>
350> **Stop conditions:** confidence <80% on any critical decision → escalate by asking the user directly · ≥3 revisions on same thought → re-frame the problem · branch count >3 → split into sub-task.
351>
352> **Implicit mode:** apply methodology internally without visible markers when adding markers would clutter the response (routine work where reasoning aids accuracy).
353>
354> **Deep-dive:** see `$sequential-thinking` skill (`.claude/skills/sequential-thinking/SKILL.md`) for worked examples (API design, debugging, architecture), advanced techniques (spiral refinement, hypothesis testing, convergence), and meta-strategies (uncertainty handling, revision cascades).
355
356<!-- /SYNC:sequential-thinking-protocol -->
357
358<!-- SYNC:critical-thinking-mindset:reminder -->
359
360**MUST ATTENTION** apply critical + sequential thinking — every claim needs appropriate traced evidence (`file:line` for repo/code claims; source URL or artifact section for research, product, content, and docs claims); confidence >80% to act, <60% DO NOT recommend. Anti-hallucination: never present guess as fact, admit uncertainty freely, cross-reference independently, stay skeptical of own confidence.
361
362<!-- /SYNC:critical-thinking-mindset:reminder -->
363
364<!-- SYNC:sequential-thinking-protocol:reminder -->
365
366**MUST ATTENTION** apply sequential-thinking — multi-step Thought N/M, REVISION/BRANCH/HYPOTHESIS markers, confidence % closer; see `$sequential-thinking` skill.
367
368<!-- /SYNC:sequential-thinking-protocol:reminder -->
369
370<!-- SYNC:ai-mistake-prevention:reminder -->
371
372**MUST ATTENTION** apply AI mistake prevention — verify generated content against evidence, trace downstream references before deleting or renaming, verify all affected outputs, re-read files after context loss, and surface ambiguity before acting.
373
374<!-- /SYNC:ai-mistake-prevention:reminder -->
375
376<!-- PROMPT-ENHANCE:STEP-TASK-CLOSING:START -->
377
378## Prompt-Enhance Closing Anchors
379
380**IMPORTANT MUST ATTENTION** follow declared step order for this skill; NEVER skip, reorder, or merge steps without explicit user approval
381**IMPORTANT MUST ATTENTION** for every step/sub-skill call: set `in_progress` before execution, set `completed` after execution
382**IMPORTANT MUST ATTENTION** every skipped step MUST include explicit reason; every completed step MUST include concise evidence
383**IMPORTANT MUST ATTENTION** if Task tools unavailable, maintain an equivalent step-by-step plan tracker with synchronized statuses
384
385<!-- PROMPT-ENHANCE:STEP-TASK-CLOSING:END -->
386
387<!-- SYNC:project-protocol-overlay -->
388
389> **Project Protocol Overlay** — Before executing this skill, resolve any PROJECT overlay rules layered onto it: match this skill's name against the `Target` column of the project's skill-protocol index (`docs/project-reference/skill-protocols-reference.md` by default; a `referenceDocs` entry in `docs/project-config.json` overrides the path), taking the most specific matching tier ONLY — exact name > glob > `*`. **That precedence orders overlays against EACH OTHER, never against this skill.** Read ONLY the matched bodies, resolved as `<protocols-dir>/<Name>.md`; a row's Body link is display text, never a read path. A matched body that is missing or malformed is REPORTED and skipped — never reconstructed from the index Description. No index, or no match -> proceed with no overlay, silently. Full contract: `.claude/skills/project-skill-protocol/references/registry.md`.
390>
391> Overlays are **ADDITIVE ONLY**: they ADD rules on top of this skill's own protocol and NEVER replace, override, disable, or reinterpret a rule it already states — removing every overlay must return this skill to exactly its documented behavior. An overlay is a BRIEF, not an authority escalation: it can NEVER waive a workflow gate, git discipline, a review gate, or a user-confirmation gate. A genuine overlay-vs-skill conflict, or two equally-specific overlays that directly contradict -> surface both to the user; NEVER resolve silently.
392
393<!-- /SYNC:project-protocol-overlay -->
394
395<!-- SYNC:project-protocol-overlay:reminder -->
396
397**MUST ATTENTION** resolve project protocol overlays for this skill BEFORE executing — most specific matching tier only (exact > glob > `*`, which ranks overlays against each other, NEVER against this skill), read only matched bodies at `<protocols-dir>/<Name>.md`; a missing or malformed body is reported, never reconstructed. Overlays are ADDITIVE ONLY (they never replace this skill's own rules) and are a brief, NEVER an authority escalation; an equal-specificity contradiction goes to the user.
398
399<!-- /SYNC:project-protocol-overlay:reminder -->
400
401## Closing Reminders
402
403**IMPORTANT MUST ATTENTION Goal:** Turn a vague product idea into a validated, tech-agnostic, module-anchored backlog artifact ready for `$refine` to convert into a PBI — preserving problem intent without leaking solution or stack choices.
404
405**IMPORTANT MUST ATTENTION — Main steps (run in order, NEVER skip/reorder):** (1) Gather info; (2) Generate artifact `IDEA-{YYMMDD}-{NNN}` draft from `idea-template.md`; (3) Capture problem/value/users; (4) Detect module via `Glob("docs/specs/*/README.md")`; (5) Load feature context (8-12K budget); (6) Save to canonical path; (6.5) Discovery Interview (ask the user directly 3-5); (7) Validate (ask the user directly 2-3); (8) Suggest next → `$refine`. — why: AI keeps dropping the skill's own mid-pipeline steps; the two gates and module detection are the most-forgotten.
406
407**IMPORTANT MUST ATTENTION — Protocols in force (concise digest of the SYNC/shared blocks this skill carries):**
408
409- **UI Wireframe:** classify each component into ONE tier; search libs first, reuse ≥80% match.
410- **AI Mistake Prevention:** verify generated content against evidence, trace downstream references, verify all affected outputs, re-read after context loss, surface ambiguity.
411- **Critical Thinking:** traced proof per claim; confidence >80% to act; never present guess as fact.
412- **Sequential Thinking:** multi-step Thought N/M with REVISION/BRANCH/HYPOTHESIS markers and confidence closer.
413
414**IMPORTANT MUST ATTENTION** Discovery Interview (Step 6.5) + Validation (Step 7) NEVER optional — run BOTH ask the user directly gates even for "simple" ideas — why: discovery uncovers hidden constraints, validation confirms problem framing; different question categories
415**IMPORTANT MUST ATTENTION** ALWAYS keep problem statement tech-agnostic (M1, `spec-principles.md` §3, all modes) — name no framework/product/language/design-pattern; defer any stack preference to the later tech-research phase — why: PBI inherits the narrative cleanly downstream
416**IMPORTANT MUST ATTENTION** in greenfield mode NEVER ask about tech stack — acknowledge a volunteered preference, then defer to the business-evaluation phase — why: stack is a research-driven decision after business analysis, not a capture-time guess
417**IMPORTANT MUST ATTENTION** task tracking break ALL work into small tasks BEFORE starting — including a task to READ `project-structure-reference.md` (skip in greenfield — it won't exist)
418**IMPORTANT MUST ATTENTION** validate all decisions with user by asking the user directly — NEVER auto-decide — and NEVER show confidence levels on an auto-detected module match
419**IMPORTANT MUST ATTENTION** auto-detect module silently via `Glob("docs/specs/*/README.md")` — prompt only when ambiguous or no match; greenfield → skip module detection — why: confirm with `Glob()` evidence, not assumption
420**IMPORTANT MUST ATTENTION** assign NO logical IDs (M3) — an idea is tech-agnostic business intent only; the downstream PBI owns `FR-`/`BR-` assignment and `[Source: namespace/service/id]` anchors — why: keep the problem/value narrative free of source identifiers so the PBI inherits it cleanly
421**IMPORTANT MUST ATTENTION** include `t_shirt_size` (XS/S/M/L/XL) in the artifact and keep the feature-context load within the 8-12K token budget — why: early sizing feeds prioritization; over-budget reads dilute attention
422**IMPORTANT MUST ATTENTION** persist to `team-artifacts/ideas/{YYMMDD}-{role}-idea-{slug}.md`, then hand off to `$refine` for PBI conversion — why: canonical path keeps downstream tooling aligned
423**IMPORTANT MUST ATTENTION** search existing component libraries before proposing any new UI component (≥80% match = reuse); classify each into exactly ONE tier — why: duplicate UI code = wrong tier
424**IMPORTANT MUST ATTENTION** cite `file:line` proof or traced evidence for every claim/recommendation, confidence >80% to act, <80% verify first — why: certainty without evidence is the root of hallucination
425**IMPORTANT MUST ATTENTION** add a final review task to verify work quality
426
427**Anti-Rationalization:**
428
429| Evasion | Rebuttal |
430| -------------------------------------------- | ----------------------------------------------------------------------- |
431| "Idea is simple, skip interview" | NEVER skip — discovery uncovers hidden constraints |
432| "Module is obvious, skip detection" | Still run `Glob()` — confirm with evidence not assumption |
433| "Validation is redundant after interview" | ALWAYS run both — different question categories |
434| "Greenfield check is optional" | Auto-detect is MANDATORY — no manual override |
435| "User mentioned a framework, capture it" | Stay tech-agnostic — acknowledge, defer to tech-research phase |
436| "I'll assign FR-/BR- IDs now" | NO logical IDs at idea stage — the PBI assigns them downstream |
437| "Reuse an existing component? new is faster" | Search libs first — ≥80% match = reuse; new without search = wrong tier |
438
439**[TASK-PLANNING]** Before acting, analyze task scope and systematically break it into small todo tasks and sub-tasks using task tracking.
440
441**IMPORTANT MUST ATTENTION** the 3 rules to never skip: (1) run BOTH Discovery + Validation ask the user directly gates; (2) keep the problem statement tech-agnostic (no stack/IDs); (3) cite `file:line` evidence, confidence >80% to act.
442
443<!-- CODEX:SYNC-PROMPT-PROTOCOLS:START -->
444
445## Hookless Prompt Protocol Mirror (Auto-Synced)
446
447Source: `.claude/.ck.json` + `.claude/skills/shared/sync-inline-versions.md` (`:full` blocks) + `.claude/scripts/lib/hookless-prompt-protocol.cjs`
448
449## [WORKFLOW-EXECUTION-PROTOCOL] [BLOCKING] Workflow Execution Protocol — MANDATORY IMPORTANT MUST CRITICAL. Do not skip for any reason.
450
451**Generic portability boundary:** Reusable skills and protocol text stay project-neutral; project-specific conventions are discovered from docs/project-config.json and docs/project-reference/. Apply shared AI-SDD from `shared/sdd-artifact-contract.md`. Read `docs/project-config.json` and `docs/project-reference/docs-index-reference.md`, then open the project reference docs named there. For spec, test-case, behavior-change, public-contract, or `docs/specs/` work, route through the local spec docs named by the docs index: `feature-spec-reference.md`, `spec-system-reference.md`, `spec-principles.md`, and `workflow-spec-test-code-cycle-reference.md` when specs/tests/code must stay synchronized. If either file or a required reference doc is missing or stale, auto-run `$project-init` (or the narrow lower-level route such as `$project-config`, `$docs-init`, `$scan-all`, or `$scan --target=<key>`) before ordinary project-specific work. Any supported AI tool may execute when this shared context and local docs are available.
452
4531. **DETECT:** If the prompt starts with an explicit slash skill/workflow command, execute it directly. Otherwise match the prompt against the workflow catalog and skill list.
4542. **ANALYZE:** Choose the best option: execute directly, invoke a skill, activate a standard workflow, or compose a custom step combination.
4553. **AUTO-SELECT:** Pick the best option yourself. Do not ask the user to choose between direct execution, skill, standard workflow, or custom workflow.
4564. **ACTIVATE:** For a selected workflow, call `$start-workflow <workflowId>`; for a selected skill, invoke that skill; for a custom workflow, sequence custom steps directly; for direct execution, proceed with the task.
4575. **CREATE TASKS:** task tracking for ALL workflow/skill/custom steps before execution when the selected path has multiple steps.
4586. **PARALLELIZE:** Before executing the task list, tag each task `PAR` (independent inputs + write set disjoint from every other `PAR` task) or `SEQ` (name the blocking dependency), group `PAR` tasks into waves, declare the wave plan, and spawn each wave's sub-agents in ONE message — all-return barrier per wave, fan-out one level deep unless a sub-agent's own definition authorizes further fan-out. Sequential-by-default is a defect when tasks are independent; do not parallelize shared write targets, output-consuming tasks, trivial single-file work, ordering a skill or workflow explicitly fixes, or user-approval gates.
4597. **EXECUTE:** Advance per the **Workflow Step Advancement & Parallel Phases** rule in your context instructions — model-driven; a sub-agent completion advances a step identically to an inline call; a parallel-phase group is an all-return barrier (advance only after ALL members return, never serialize it)
460
461## Shared AI-SDD Protocol Markers
462
463Source: `.claude/skills/shared/sync-inline-versions.md`
464
465## SYNC:ai-sdd-artifact-contract
466
467> **AI-SDD Artifact Contract** — Shared spec-driven development rules stay portable and source-owned.
468>
469> 1. Keep reusable AI-SDD principles in `.claude`; put repository-specific paths, commands, owners, products, and formats in project config/reference docs.
470> 2. Preserve cycle: `spec -> plan -> tasks -> implement -> verify -> update spec/docs`.
471> 3. Trace every requirement or invariant through decision, task, TC/test, source evidence, and docs/spec update.
472> 4. Treat code-to-spec extraction as reference-only until accepted by the canonical spec owner.
473> 5. Any supported AI tool may plan, implement, review, or verify with synced context; using multiple tools is optional.
474> 6. Update `.claude` source first, then sync generated mirrors; do not manually edit `.agents`, `.codex`, or `AGENTS.md`. — why: mirrors are generated artifacts; hand-edits are overwritten on the next sync
475> 7. If `docs/project-config.json`, root instruction files, or a required project-reference doc is missing or stale, auto-run `$project-init` or the narrow lower-level route before ordinary project-specific work.
476>
477> **Active reference:** `shared/sdd-artifact-contract.md` in the active skills root.
478
479---
480
481## SYNC:ai-sdd-artifact-contract:reminder
482
483- **MANDATORY** Apply `shared/sdd-artifact-contract.md`; keep reusable AI-SDD in `.claude` and local rules in project docs.
484- **MANDATORY** Code-to-spec extraction is reference-only until canonical acceptance; any supported AI tool may execute with synced context.
485- **MANDATORY** Update `.claude` source before syncing generated mirrors; do not manually edit `.agents`, `.codex`, or `AGENTS.md`.
486- **MANDATORY** Missing or stale project config, root instruction files, or required reference docs route project-specific work through `$project-init` or the narrow setup route automatically.
487 **[TASK-PLANNING] [MANDATORY]** BEFORE executing any workflow or skill step, create/update task tracking for all planned steps, then keep it synchronized as each step starts/completes.
488
489## [LESSON-LEARNED-REMINDER] [BLOCKING] Task Planning & Continuous Improvement — MANDATORY. Do not skip.
490
491Break work into small tasks (task tracking) before starting. Add final task: "Analyze AI mistakes & lessons learned".
492
493**Extract lessons — ROOT CAUSE ONLY, not symptom fixes:**
494
4951. Name the FAILURE MODE (reasoning/assumption failure), not symptom — "assumed API existed without reading source" not "used wrong enum value".
4962. Generality test: does this failure mode apply to ≥3 contexts/codebases? If not, abstract one level up.
4973. Write as a universal rule — strip project-specific names/paths/classes. Useful on any codebase.
4984. Consolidate: multiple mistakes sharing one failure mode → ONE lesson.
4995. **Recurrence gate:** "Would this recur in future session WITHOUT this reminder?" — No → skip `$learn`.
5006. **Auto-fix gate:** "Could `$code-review`/`$code-simplifier`/`$security-review`/`$lint` catch this?" — Yes → improve review skill instead.
5017. BOTH gates pass → ask user to run `$learn`.
502 **[CRITICAL-THINKING-MINDSET]** Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence >80% to act.
503 **Anti-hallucination principle:** Never present guess as fact — cite sources for every claim, admit uncertainty freely, self-check output for errors, cross-reference independently, stay skeptical of own confidence — certainty without evidence root of all hallucination.
504 **AI Attention principle (Primacy-Recency):** Put the 3 most critical rules at both top and bottom of long prompts/protocols so instruction adherence survives long context windows.
505 **Goal-driven execution:** Define success criteria first, loop until verified, and stop only when observable checks pass.
506 **Tests verify intent:** Tests must protect business rules/invariants and fail when the protected intent breaks, not only mirror current behavior.
507
508## Common AI Mistake Prevention (System Lessons)
509
510- **Re-read files after context compaction.** Edit requires prior Read in same context; compaction wipes read state. Re-read before editing.
511- **Grep for old terms after bulk replacements.** AI over-trusts find/replace completeness. Grep full repo after bulk edits for missed refs in docs/configs/catalogs.
512- **Check downstream references before deleting.** Deletions cascade doc/code staleness. Map referencing files before removal.
513- **After memory loss, check existing state before creating new.** Compaction wipes prior-work memory. Query current state to resume — never blindly duplicate.
514- **Verify AI-generated content against actual code.** AI hallucinates APIs, class names, method signatures. Grep to confirm existence before documenting/referencing.
515- **Trace full dependency chain after edits.** Changing a definition misses downstream consumers. Trace the full chain.
516- **When renaming, grep ALL consumer file types.** Some file types silently ignore missing refs (no compile error). Search code, templates, configs, generated files.
517- **Trace ALL code paths when verifying correctness.** Code existing ≠ code executing. Trace early exits, error branches, conditional skips — not just happy path.
518- **Update docs that embed canonical data when source changes.** Docs inlining derived data (workflows, schemas, configs) go stale silently. Update all embedding docs alongside source.
519- **Verify sub-agent results after context recovery.** Background agents may finish while parent compacted — grep-verify output, don't trust assumed completion.
520- **Cross-check full target list against sub-agent assignments.** Parallel sub-agents by category miss boundary items. Reconcile union of assignments against target list before proceeding.
521- **Sub-agents inherit knowledge only from their agent .md definition — use custom agent types, not built-in Explore.** Tool adoption = permission + knowledge + enforcement (numbered workflow step).
522- **Persist sub-agent findings incrementally, not as a final batch.** Long sub-agents hit cutoffs before final write — findings lost. Instruct append-per-section to report file.
523- **When debugging, ask "whose responsibility?" before fixing.** Trace caller (wrong data) vs callee (wrong handling). Fix at responsible layer — never patch symptom site.
524- **Test failure → record a provisional verdict before trace/edit, then investigate.** Use the full five-way taxonomy: SOURCE-WRONG (production violates intent), TEST-WRONG (assertion/setup is stale), TEST-NOT-OPTIMAL (valid but fragile or low-signal test), ENVIRONMENT-BLOCKED (external state prevents a verdict), or AMBIGUOUS (intent/evidence cannot choose safely). Then trace root cause and triangulate against the governing spec (`docs/specs/**` if one exists) AND source. NEVER weaken an assertion, add a skip, relax a timeout, or change source merely to force green.
525- **Grep ALL removed names after extraction/refactoring.** Primary file "done" ≠ secondary files clean. Grep entire scope for every removed symbol before declaring complete.
526- **Assume existing values are intentional — ask WHY before changing OR flagging one as a defect.** Pattern-matching as "wrong" skips context. Before changing or reporting any constant/limit/flag/cutoff: read comments, git blame, the CALLER's ordering (the guarantee that makes the value correct usually lives in code running immediately BEFORE the cited line), and 2+ sibling call sites of the same convention. A doc stating WHAT without WHY is missing rationale, not proof of a missing guard — and in a validation pass, an accurate `file:line` citation proves the transcription, never the defect.
527- **Verify ALL affected outputs, not just the first.** One build green ≠ all green. Multi-stack changes (backend/frontend/tests/docs) require verifying EVERY output.
528- **Evaluate fit before copying a nearby pattern.** Closest example ≠ matching preconditions — verify the new context shares the same constraints, base classes, scope, lifetime.
529- **Holistic-first debugging — resist nearest-attention trap.** Don't dive into first plausible cause. List EVERY precondition (config, env vars, paths, DB, endpoints, creds, versions, DI, data). Verify each against evidence (grep/query — not reasoning). Ask "what would falsify this?" — if nothing, it's not a hypothesis. Most expensive failure: going deeper in "obvious" layer while bug sits in layer never questioned.
530- **Surgical changes — apply the diff test (context-aware).** Two modes: (1) Bug fix → every line traces to the bug; no restyling; orphan cleanup only for imports YOUR changes made unused. (2) Review/enhancement → implement improvements AND announce as "Enhancement beyond main request: [what]". Never silently scope-creep. Diff test: "Would this line exist if I wasn't asked to do X?" — if no, delete or announce.
531- **Surface ambiguity before coding — don't pick silently.** Multiple valid interpretations → present each with effort: "[Request] could mean (1) [N h], (2) [N h]. Which matters?" List scope/format/volume/constraints assumptions first. If simpler path exists, say so. Never silently pick.
532- **[MANDATORY FIRST ACTION] ALWAYS activate a suitable skill or workflow BEFORE responding.** Match task against workflow catalog + skill list; invoke via skill invocation or `$start-workflow <workflowId>`. NEVER answer or write code before checking. Skip = protocol violation.
533- **Why-Review adversarial mindset — apply when reviewing any plan, decision, or design.** Default SKEPTIC not VALIDATOR: steel-man a rejected alternative, invert each stated reason ("what does it sacrifice?"), stress-test top 2-3 assumptions, run pre-mortem ("ships, fails in 3 months — what breaks?"), surface 1-2 alternatives author missed. Section presence ≠ quality; quality = causal reasoning + concrete mitigations + evidence, not "it's better" or "monitor closely".
534- **Front-load report-write in sub-agent prompts for large reviews.** Many-file sub-agents hit budget before final write — findings lost. Design prompts so: (1) report-write is first explicit deliverable, (2) append per-file/section (not batched), (3) scope bounded so reads don't exhaust budget. Truncated mid-sentence with no report file → spawn narrower scope, don't retry same prompt.
535- **After context compaction, re-verify all prior phase outcomes before continuing.** Summaries describe intent, not environment state (git index, filesystem, processes). On resume, FIRST audit: git status, re-read modified files, verify filesystem. Every "completed" claim is an untested hypothesis until evidence confirms.
536- **OOM/memory: check row count before row size.** Triage: (1) Unbounded query — no DB filter for trigger? Push filter to DB; eliminates OOM. (2) Large rows? Projection reduces proportionally. Row reduction > projection in ROI.
537- **Assert the outcome your system OWNS, never the intermediate state your INFRASTRUCTURE owns.** When testing anything asynchronous (queue/broker delivery, retries, background jobs, caches, replication), assert the final business/entity state. NEVER assert the delivery bookkeeping — consume/send status, attempt counts, last-error, row existence or counts in a broker, scheduler, or outbox/inbox table. That bookkeeping lives in shared infrastructure that ANY co-running process (a peer worker, a second replica, a leftover local container) can write, usually under a deterministic shared key, so the assertion silently tests the developer's environment instead of the system: green when run alone, flaky the instant anything else shares that broker + database. Gate question for every assertion: "would this hold no matter WHICH process did the work?" — if no, assert the converged data state instead. Corollary: process-local fault injection and in-process telemetry cannot gate work any process may perform — use them as stress amplifiers (arm → bounded window → disarm → assert convergence), never as preconditions.
538- **Keep domain concepts out of generic/shared/infrastructure layers.** Reusable layer (shared library, framework, infra module) must reference NO consumer-specific domain concept — tenant/customer/product IDs, business entities, feature rules. Leak compiles + runs → passes review silently while coupling the "reusable" layer to one consumer. Keep shared type domain-free; push domain fields/logic down into the consumer via subclass/composition. — why: a layer coupled to one consumer's domain is no longer reusable.
539
540<!-- CODEX:SYNC-PROMPT-PROTOCOLS:END -->