Project Config

Quick Summary

Goal: Scan workspace, update docs/project-config.json with accurate values.

IMPORTANT MUST ATTENTION follow Plan → Review → Execute workflow. IMPORTANT MUST ATTENTION use exact schema field names (--describe). IMPORTANT MUST ATTENTION validate after each phase. NEVER use classPattern/keyExtractor — correct fields: contentPattern/keyGroup.

Workflow: Recon → classify scale → /plan → /plan-review → Execute phases (scan → merge → validate → fix) → Follow-up scans → /prompt-enhance

Key Rules:

• MUST ATTENTION run node .claude/hooks/lib/project-config-schema.cjs --describe — use field names verbatim
• MUST ATTENTION execute every required config section for every project size; small projects do not skip, defer, or require user approval to combine work
• MUST ATTENTION one TaskCreate per config section or explicit section group — NEVER scan everything in one pass
• MUST ATTENTION validate schema after each merge — validateConfig(config) returns PASSED or errors
• MUST ATTENTION review-and-fix after each phase — read back, spot-check paths, self-review
• MUST ATTENTION do not ask the user to choose scan granularity, combination, section ordering, or optional confirmation; auto-select the evidence-backed route and continue
• Path regexes MUST ATTENTION use [\\/] for cross-OS separator matching
• Schema enforced by .claude/hooks/lib/project-config-schema.cjs

⛔ Plan → Review → Execute Workflow

Step 1: Detect — Classify Project Scale

MUST ATTENTION classify scale FIRST — drives task granularity for all subsequent phases.

find . -path "*/node_modules" -prune -o -name "*.csproj" -print 2>/dev/null | wc -l
find . -path "*/node_modules" -prune -o -name "package.json" -print 2>/dev/null | wc -l
find . -path "*/node_modules" -prune -o -type f -name "{configured-source-file-glob}" -print 2>/dev/null | wc -l
find . -maxdepth 3 -type d -name "{candidate-source-dir-name}" 2>/dev/null

Project size controls task grouping and split depth only. It does NOT permit skipping required sections, stopping for user approval, or asking whether to combine work. For small projects, auto-select the compact full-pass plan and keep validating after each merge/review phase.

Step 2: Create Plan (/plan)

Create plans/{date}-project-config-scan.md:

1. Record scale classification from Step 1
2. Group config sections into phases (≤5 tasks each) while preserving full section coverage
3. Include review-and-fix cycle after each phase
4. Include every Phase 2 section (2a–2q) as either its own task or a named task inside a compact group with explicit evidence for each section

Phase template:

Phase A: Setup & Metadata — validate config, read schema, scan project metadata
Phase B: Module Discovery — backend projects, frontend apps/libs, framework keywords
Phase C: Context & UI — context groups, design system, styling, component system
Phase D: Testing & Infra — testing, E2E, databases, messaging, API, infrastructure
Phase E: Graph Connectors — API endpoints, implicit connections, referenceDocs
Phase F: Final Review — consolidate, validate, hook tests, create /scan-* tasks
Phase G: Self-Review — re-invoke /project-config to verify all config matches source code

Step 3: Review Plan (/plan-review)

Run /plan-review on the generated scan plan; resolve blocking findings before executing any phase.

Step 4: Execute

Per phase: TaskCreate → scan → merge → validate → spot-check → fix → next phase.

Review-and-Fix Cycle (MANDATORY per phase)

1. Read back updated config sections
2. Spot-check 2–3 paths against actual dirs
3. Run schema validation
4. Self-review: missing modules? Accurate descriptions? Correct regexes?
5. Fix before proceeding

Intermediate Workspace

Medium/large projects: mkdir -p.ai/workspace/project-config — write phase reports before merging. Delete after consolidation.

⛔ Schema Protection Rules

NEVER rename/remove/restructure top-level sections. NEVER change field types. NEVER populate deprecated v1 sections for new projects. NEVER remove v1 data from existing projects.

MAY add entries to maps/arrays, update values, add optional schema fields, populate v2 sections.

Schema Structure (v2)

docs/project-config.json
├── schemaVersion, project{ name, description, languages[], packageManagers[], monorepoTool }
├── modules[] — { name, kind, pathRegex, description, tags[], meta{} }
├── contextGroups[] — { name, pathRegexes[], fileExtensions[], guideDoc, patternsDoc, stylingDoc, designSystemDoc, rules[] }
├── styling — { technology, guideDoc, appMap{}, patterns[] }
├── designSystem — { docsPath, modernUiNote, appMappings[] }
├── componentSystem — { type, selectorPrefixes[], filePattern, layerClassification{} }
├── framework — { name, backendPatternsDoc, frontendPatternsDoc, codeReviewDoc, integrationTestDoc, searchPatternKeywords[] }
├── testing — { frameworks[], filePatterns{}, commands{}, coverageTool, guideDoc, integrationRules[] }
├── e2eTesting — { framework, language, configFile, testsPath, pageObjectsPath, fixturesPath, ... }
├── databases{}, messaging{ broker, patterns[], consumerConvention }, api{ style, docsFormat, docsPath, authPattern }
├── infrastructure — { containerization, orchestration, cicd{ tool, configPath } }
├── graphConnectors — apiEndpoints{ enabled, frontend{ framework, paths[] }, backend{ framework, paths[], routePrefix } }
│   └── implicitConnections[] — { name, edgeKind, paths[], source{ filePattern, contentPattern, keyGroup }, target{...}, matchBy }
├── referenceDocs[] — { filename, purpose, sections[] }
├── integrationTestVerify — { guidance, referenceDocs[], quickRunCommand, testProjectPattern, testProjects[], systemCheckCommand, runScript, startupScript }
├── workflowPatterns — { architectureStyle, codeHierarchy, cssMethodology, stateManagement, crossModuleValidation, featureDocTemplate, reviewRulesDoc }
├── specRoots — { business{ path, authorship, m1Policy }, technical{ path, authorship, m1Policy } }  (drives /spec + /tech-spec)
├── techSpecScan — { sourceRoot, fileExtensions[], annotationPattern }  (enables /tech-spec) | else _techSpecScanNote (deliberate-omission carrier)
└── DEPRECATED: backendServices, frontendApps, scss, componentFinder, sharedNamespace

MUST ATTENTION run node .claude/hooks/lib/project-config-schema.cjs --describe for exact field names.

⛔ Common AI Field Name Mistakes

Deriving the Spec-System Config From Source (specRoots + techSpecScan)

These two sections drive /spec (hand-authored business specs) and /tech-spec (derived technical specs). They are optional in the schema but effectively required for any project that wants /tech-spec — the generator exits non-zero when techSpecScan is absent. Derive each value from THIS project's own source; never copy another project's literals. --describe (Phase 0b) now emits a # derivation note per field — read it.

specRoots — declare each tree's location AND semantics:

techSpecScan — how /tech-spec finds annotated tests. Derive all three from the stack:

1. sourceRoot — the project's primary source dir (from project. layout / modules[].pathRegex), e.g. src.

2. fileExtensions — the source-language extensions (from project.languages), e.g. [".cs"], [".ts"], [".java"].

3. annotationPattern — a regex matching THIS project's spec-annotation convention. CONTRACT (hard): exactly two capture groups — group 1 = trait name (TestSpec or TechnicalSpec), group 2 = the non-empty spec id. The generator validates both at parse time and refuses to run on a mismatch. Detect the convention by grepping existing tests, then build the pattern:

   Stack / convention (illustrative)	Example annotation in source	annotationPattern
   C# xUnit [Trait]	[Trait("TestSpec", "TC-001")]	\[Trait\("(TestSpec|TechnicalSpec)"\s*,\s*"([^"]+)"\)\]
   Java/Kotlin JUnit @Tag	@Tag("TestSpec:TC-001")	@Tag\("(TestSpec|TechnicalSpec):([^"]+)"\)
   TS/JS test title tag	describe('[TestSpec:TC-001] ...')	\[(TestSpec|TechnicalSpec):([^\]]+)\]

   Adapt to whatever the project actually uses — the table is illustrative, not a fixed list. Validate the 2-group contract before saving:

   node -e "const p=/YOUR_PATTERN/; const c=(new RegExp(p.source+'|')).exec('').length-1; console.log('capture groups:', c, c===2?'OK':'FIX — must be exactly 2 (trait, id)')"
   

4. No spec-annotation convention? Do NOT invent one and do NOT leave techSpecScan half-filled. Omit techSpecScan and set _techSpecScanNote with a one-line reason — the generator then exits cleanly naming the missing key instead of reporting zero annotations, and the omission reads as a decision.

Phase 0: Setup

# 0a. Validate current config
node -e "const{validateConfig,formatResult}=require('./.claude/hooks/lib/project-config-schema.cjs');const c=JSON.parse(require('fs').readFileSync('docs/project-config.json','utf-8'));console.log(formatResult(validateConfig(c)))"

# 0b. Read exact schema shapes (MANDATORY)
node .claude/hooks/lib/project-config-schema.cjs --describe

# 0c. Check CLAUDE.md
test -f CLAUDE.md && echo "EXISTS" || echo "MISSING"

# 0d. Create workspace
mkdir -p .ai/workspace/project-config

Phase 1: Read Current Config

Read docs/project-config.json. Note populated vs skeleton sections.

Phase 2: Section-by-Section Scans

Each subsection = one TaskCreate or an explicit named child inside a compact group. Per task: investigate → report → merge → validate. Small projects still cover every subsection; compact grouping is an execution convenience, not permission to skip or ask.

2a. Modules — Backend

find . -path "*/node_modules" -prune -o -name "*.csproj" -print | head -50
find . -name "pom.xml" -o -name "build.gradle" | head -50
find . -path "*/node_modules" -prune -o -name "package.json" -print | head -50
find . -name "go.mod" | head -50

Build modules[] entries: { name, kind, pathRegex, description, tags[], meta{} }

• kind: "backend-service", "library", "framework"

2b. Modules — Frontend

find . -name "nx.json" -o -name "{frontend-framework-config}" -o -name "lerna.json" -o -name "turbo.json" 2>/dev/null | head -5
find . -maxdepth 5 -type d \( -name apps -o -name libs -o -name packages \) 2>/dev/null | head -30

Build entries: kind: "frontend-app" or kind: "library".

2c. Project Metadata

Detect languages (.cs→csharp, .ts→typescript, .py→python, .java→java, .go→go), package managers, monorepo tool. Build project { name, description, languages[], packageManagers[], monorepoTool }.

2d. Framework Patterns

Grep abstract class, interface I, most-imported symbols. Build framework { name, searchPatternKeywords[] } from commonly used base classes.

2e. Context Groups

Build contextGroups[] with pathRegexes[], fileExtensions[], patternsDoc, rules[]. Rules MUST ATTENTION be specific: "Use the service-specific repository (e.g. OrderRepository), not the generic repository base" not "follow best practices".

2f–2h. Design System, Styling, Component System

• designSystem { docsPath, modernUiNote, appMappings[] }
• styling { technology, fileExtensions, guideDoc, appMap{}, patterns[] }
• componentSystem { type, selectorPrefixes[], filePattern, layerClassification{} }

2i–2j. Testing & E2E

• testing { frameworks[], filePatterns{}, commands{}, coverageTool, guideDoc, integrationRules[] }
• e2eTesting { framework, language, configFile, testsPath, pageObjectsPath, fixturesPath, runCommands{}, tcCodeFormat, entryPoints[] }
• integrationTestVerify { guidance, referenceDocs[], runScript, startupScript, quickRunCommand, systemCheckCommand, testProjectPattern, testProjects[] }
• integrationTestVerify.referenceDocs[] MUST contain project-specific docs that explain setup prerequisites before a verifier runs systemCheckCommand or test commands.

2k–2n. Databases, Messaging, API, Infrastructure

• databases {} (freeform)
• messaging { broker, patterns[], consumerConvention }
• api { style, docsFormat, docsPath, authPattern }
• infrastructure { containerization, orchestration, cicd{ tool, configPath } }

2o. Graph Connectors — API Endpoints

Only if project has BOTH frontend AND backend.

Route prefix: derive from configured backend framework and existing route declarations.

2p. Graph Connectors — Implicit Connections

⛔ How implicitConnections Works (MUST ATTENTION UNDERSTAND)

Algorithm: scan source files → extract keys via contentPattern regex capture group keyGroup → scan target files → match keys via matchBy → create edgeKind edges.

Exact Schema Fields

source/target fields: filePattern (glob, e.g. "*.cs"), contentPattern (regex WITH capture group), keyGroup (1-based integer, default 1)

⛔ NEVER use classPattern, keyExtractor, pattern. ALWAYS use contentPattern, keyGroup.

Detection Heuristics

• Configured runtime: discover event, handler, publisher, and consumer base types from codebase grep and project-reference docs.
• TypeScript: Redux dispatch→reducer, NgRx createAction→ofType, EventEmitter emit→on
• Python: Celery task.delay→@app.task, Django signal.send→@receiver
• Java: publishEvent→@EventListener, KafkaTemplate→@KafkaListener

Example (correct format)

{
    "name": "entity-to-event-handlers",
    "edgeKind": "MESSAGE_BUS",
    "paths": ["{configured-domain-source-root}/", "{configured-application-source-root}/{event-handler-folder}/"],
    "source": { "filePattern": "{configured-source-file-glob}", "contentPattern": "{configured-entity-pattern}", "keyGroup": 1 },
    "target": { "filePattern": "{configured-source-file-glob}", "contentPattern": "{configured-event-handler-pattern}", "keyGroup": 1 },
    "matchBy": "key-contains"
}

IMPORTANT MUST ATTENTION record detected rules in the plan/report before writing; do not pause for user approval. IMPORTANT MUST ATTENTION scope paths to relevant dirs (not repo root).

2q. Reference Docs — Canonical Floor (MUST normalize, NEVER raw-import)

⛔ Reference docs are the FRAMEWORK's canonical set, not whatever files happen to sit in docs/project-reference/. Do NOT build referenceDocs[] by listing on-disk files — that silently re-imports drift (legacy filenames like feature-docs-reference.md, missing canonical docs, wrong order). Normalize against the canonical floor instead:

node -e "const h=require('./.claude/hooks/lib/session-init-helpers.cjs');const{loadProjectConfig}=require('./.claude/hooks/lib/project-config-loader.cjs');console.log(JSON.stringify(h.normalizeReferenceDocs((loadProjectConfig()||{}).referenceDocs),null,2))"

• Set config.referenceDocs = the returned normalized array (canonical docs + genuine project-specific extras, canonical order, legacy names resolved, canonical templatePaths preserved). Add project-specific reference docs only as EXTRA entries; never delete or rename a canonical entry.
• For each renames[] {from,to}: if docs/project-reference/<from> exists — git mv it to <to> when <to> is absent; if <to> already exists, <from> is a stale duplicate → confirm <to> holds the canonical content, then git rm <from>. Migrate every downstream textual reference (docs-index-reference.md, project-structure-reference.md) <from> → <to>.
• added[] are canonical docs missing on disk — the SessionStart hook (or the matching /scan --target=<key>) creates them. Do not hand-fabricate content; per-doc purpose/sections come from DEFAULT_REFERENCE_DOCS.
• Re-run the probe after merging; changed:false with empty renames/added/removedLegacy is the only PASS state.

Phase 3: Consolidate & Write

Merge section-by-section. Overwrite only with concrete scan findings. Large projects: merge incrementally.

Phase 4: Verify (MANDATORY)

1. Schema validation — MUST ATTENTION pass with zero errors
2. Spot-check 2–3 service paths — verify each path exists (file:line evidence)
3. Run hook tests: node .claude/hooks/tests/test-all-hooks.cjs

Phase 5: Follow-Up Tasks

Then: /claude-md-init (LAST). Optionally: /graph-build.

Phase 6: Enhance Generated Docs (MANDATORY)

Run /prompt-enhance on all generated/updated docs and CLAUDE.md. One task per file, parallel OK.

Phase 7: Self-Review Verification (MANDATORY)

Re-invoke skill: /project-config Self review and verify everything again, ensure all is correct with current source code. Catches regressions and issues missed in first pass.

Output

Report: sections updated vs unchanged, new modules discovered, path mismatches, follow-up tasks created. Include the project scale, the selected full-coverage task grouping, and confirmation that no required section was skipped because the project was small.

[IMPORTANT] Use TaskCreate to break ALL work into small tasks BEFORE starting.

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.

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 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.

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: Scan workspace, update docs/project-config.json with accurate, schema-valid values via Plan → Review → Execute.

Protocols in force (concise digest of the SYNC/shared blocks this skill carries):

• Critical Thinking: apply critical + sequential thinking; trace every claim, confidence >80% to act.
• AI Mistake Prevention: verify generated content against evidence, trace downstream references, verify all affected outputs, re-read after context loss, surface ambiguity.

IMPORTANT MUST ATTENTION classify project scale FIRST (Step 1) — drives all task granularity decisions. IMPORTANT MUST ATTENTION plan first — recon → /plan → /plan-review → execute. NEVER jump to scanning. IMPORTANT MUST ATTENTION execute all required sections for all project sizes; small projects get compact full-coverage grouping, never a permission question or skipped sections. IMPORTANT MUST ATTENTION break into phases with review cycles — scan → merge → validate → spot-check → fix per phase. IMPORTANT MUST ATTENTION use exact schema field names — run --describe, copy verbatim. NEVER guess. IMPORTANT MUST ATTENTION validate after EACH phase — schema errors compound across phases. NEVER use classPattern/keyExtractor — correct fields: contentPattern (regex) + keyGroup (number). IMPORTANT MUST ATTENTION one TaskCreate per config section — NEVER monolithic scan. IMPORTANT MUST ATTENTION Phase 7 self-review is MANDATORY — catches what every earlier phase missed.

Anti-Rationalization:

[TASK-PLANNING] Before acting, analyze task scope and systematically break it into small todo tasks and sub-tasks using TaskCreate.