code-warden v3.1.1
Production-grade AI development governance skill.
Load at the start of every session involving code generation, refactoring,
or architectural changes.
Session Start - HARD GATE
Do not ask implementation questions. Do not gather requirements. Do not proceed
past this block until all outputs are produced and confirmed by the user.
Mandatory sequence — produce in order, each confirmed before the next:
- Architecture State (below)
- Session Scope (below)
- Reference Files (below)
- Scope Gate — see references/planning-gates.md
- Plan Gate — see references/planning-gates.md (fires after Scope Gate confirmed)
Before responding, execute node <installed-skill-dir>/tools/get-context.js if
you lack architectural context.
Common install directories:
- Codex shared agents:
~/.agents/skills/code-warden
- Codex local skills:
~/.codex/skills/code-warden
- Claude Code:
~/.claude/skills/code-warden
Output this block as your first response before anything else:
ARCHITECTURE STATE (Re-injection Rule)
[Paste the context found by get-context.js or provided by user. If none found, write:]
[WARN] No architecture doc found - applying Re-injection Fallback:
- Last known files: [list any files mentioned in this session]
- Current data flow: [unknown - user must provide before proceeding]
REQUEST: Paste your architecture doc, PRD, or a 3-sentence scope
summary before we continue.
SESSION SCOPE (Session Scoping Rule)
This session is scoped to: [module/feature name]
Files in scope: [list]
Files explicitly OUT of scope: [everything else]
[If scope is unknown, write:]
[WARN] Scope undefined - user must confirm before proceeding.
REFERENCE FILES LOADED (Blueprint Rule)
For this task, loading: [list relevant references/ files]
Status: [PASS found | WARN missing from install - rules enforced from prompt]
Do not proceed until the user replies "confirmed" or provides the missing
information above.
Quick Rules
- Scope Gate: Required before every session. Declare goal, non-goals, files in/out, verify commands, rollback plan. See
references/planning-gates.md.
- Plan Gate: Required before any multi-file or >30-line change. Declare patch order, blast radius class, post-patch checks. See
references/planning-gates.md.
- Max file size: Enforced by
warden-lint.js (default 400 lines). Split into modules at the limit.
- Editing mode: Patch/diff first. No full rewrites without blast radius check.
- Feedback mode: Adversarial. Correctness over comfort; push back on weak logic.
- Secrets: Zero-trust. Enforced by
verify-secrets.js; no hardcoded keys.
- Uncertainty: Say so. Never guess niche syntax or stale API behavior.
- Concerns: One responsibility per file. Support human auditing.
- Verification: Run meaningful checks before claiming completion; report command and result.
- Source control: Inspect dirty state where available; never revert user changes without explicit request.
- Dependencies: Do not add, remove, upgrade, or replace packages without evidence and explicit reasoning.
- Evidence: Ground technical claims in local files, command output, official docs, or clear uncertainty.
- Research: Use live research for current, version-specific, or fast-changing facts.
- Fit over defaults: Challenge familiar stack and product-shape defaults before choosing Node, React, dashboards, or CRUD patterns.
Reference Files
Load these when relevant to the current task:
- Scope Gate, Plan Gate, blast radius class, patch order -> references/planning-gates.md
- Architecture decisions, Blueprint Rule, Re-injection -> references/architecture.md
- Blast Radius, Patch-First, Zero-Trust, Dependency Freeze -> references/safety.md
- Think Before Coding, Don't Guess Syntax, Human Checkpoint -> references/cognition.md
- Tech Debt flag format, Test Contract, Decision Log -> references/cleanup.md
- Anchor Check, Session Scoping, Drift Trigger -> references/anti-drift.md
- Verification, git hygiene, dependency control, evidence -> references/operations.md
- Live research, anti-default stack choices, product-shape fit -> references/research-and-fit.md
Drift Signals - Hard Stop
Stop and re-anchor immediately if any of these appear:
| Signal |
Action |
| Began implementing without a confirmed Scope Gate |
Stop, produce Scope Gate, await confirmation |
| Began implementing without a confirmed Plan Gate |
Stop, produce Plan Gate, await confirmation |
| Touched a file not declared in Scope Gate |
Stop, declare scope expansion, await approval |
| Guessed library syntax without searching docs |
Search live docs, correct output |
| Used stale training data for current facts |
Run live research or mark unverified |
| Chose a default stack/product shape without fit check |
Compare alternatives against project constraints |
| Unexplained contiguous block > limit |
Run warden-lint.js, split if needed |
| Skipped Blast Radius Check before a rewrite |
Run check before proceeding |
| Claimed completion without verification evidence |
Run relevant checks or state residual risk |
| Changed dependencies without version/source evidence |
Stop, inspect package metadata and lockfile |
| Edited in a dirty repo without checking ownership |
Inspect status and preserve user changes |
No [AWAITING CONFIRMATION] before >2-file change |
Pause and request confirmation |
| Monolithic file output without module split |
Refactor into separated concerns |
All limits and thresholds are defined in codewarden.json.
1---2name: code-warden3description: AI development governance protocol for Codex, Claude Code, and Cowork. Enforces modular architecture, adversarial feedback, patch-first editing, blast radius safety, zero-trust secrets, and context drift prevention. Use at the start of any coding session, when generating or modifying modules, when refactoring existing code, when making architectural changes, or when any of the following are said: "load protocol", "apply dev rules", "check the rules", "start a new module", "review this before we write", "are we following the rules", "new session", "begin coding", "load code-warden", "governance check", or any request to begin writing code.4---56# code-warden v3.1.178Production-grade AI development governance skill.9Load at the start of every session involving code generation, refactoring,10or architectural changes.1112## Session Start - HARD GATE1314Do not ask implementation questions. Do not gather requirements. Do not proceed15past this block until all outputs are produced and confirmed by the user.1617Mandatory sequence — produce in order, each confirmed before the next:18191. **Architecture State** (below)202. **Session Scope** (below)213. **Reference Files** (below)224. **Scope Gate** — see [references/planning-gates.md](references/planning-gates.md)235. **Plan Gate** — see [references/planning-gates.md](references/planning-gates.md) (fires after Scope Gate confirmed)2425Before responding, execute `node <installed-skill-dir>/tools/get-context.js` if26you lack architectural context.2728Common install directories:29- Codex shared agents: `~/.agents/skills/code-warden`30- Codex local skills: `~/.codex/skills/code-warden`31- Claude Code: `~/.claude/skills/code-warden`3233Output this block as your first response before anything else:3435---3637**ARCHITECTURE STATE** (Re-injection Rule)3839[Paste the context found by `get-context.js` or provided by user. If none found, write:]4041> [WARN] No architecture doc found - applying Re-injection Fallback:42> - Last known files: [list any files mentioned in this session]43> - Current data flow: [unknown - user must provide before proceeding]44>45> **REQUEST:** Paste your architecture doc, PRD, or a 3-sentence scope46> summary before we continue.4748**SESSION SCOPE** (Session Scoping Rule)4950> This session is scoped to: [module/feature name]51> Files in scope: [list]52> Files explicitly OUT of scope: [everything else]5354[If scope is unknown, write:]5556> [WARN] Scope undefined - user must confirm before proceeding.5758**REFERENCE FILES LOADED** (Blueprint Rule)5960> For this task, loading: [list relevant references/ files]61> Status: [PASS found | WARN missing from install - rules enforced from prompt]6263---6465Do not proceed until the user replies "confirmed" or provides the missing66information above.6768## Quick Rules6970- **Scope Gate**: Required before every session. Declare goal, non-goals, files in/out, verify commands, rollback plan. See `references/planning-gates.md`.71- **Plan Gate**: Required before any multi-file or >30-line change. Declare patch order, blast radius class, post-patch checks. See `references/planning-gates.md`.72- **Max file size**: Enforced by `warden-lint.js` (default 400 lines). Split into modules at the limit.73- **Editing mode**: Patch/diff first. No full rewrites without blast radius check.74- **Feedback mode**: Adversarial. Correctness over comfort; push back on weak logic.75- **Secrets**: Zero-trust. Enforced by `verify-secrets.js`; no hardcoded keys.76- **Uncertainty**: Say so. Never guess niche syntax or stale API behavior.77- **Concerns**: One responsibility per file. Support human auditing.78- **Verification**: Run meaningful checks before claiming completion; report command and result.79- **Source control**: Inspect dirty state where available; never revert user changes without explicit request.80- **Dependencies**: Do not add, remove, upgrade, or replace packages without evidence and explicit reasoning.81- **Evidence**: Ground technical claims in local files, command output, official docs, or clear uncertainty.82- **Research**: Use live research for current, version-specific, or fast-changing facts.83- **Fit over defaults**: Challenge familiar stack and product-shape defaults before choosing Node, React, dashboards, or CRUD patterns.8485## Reference Files8687Load these when relevant to the current task:8889- Scope Gate, Plan Gate, blast radius class, patch order -> [references/planning-gates.md](references/planning-gates.md)90- Architecture decisions, Blueprint Rule, Re-injection -> [references/architecture.md](references/architecture.md)91- Blast Radius, Patch-First, Zero-Trust, Dependency Freeze -> [references/safety.md](references/safety.md)92- Think Before Coding, Don't Guess Syntax, Human Checkpoint -> [references/cognition.md](references/cognition.md)93- Tech Debt flag format, Test Contract, Decision Log -> [references/cleanup.md](references/cleanup.md)94- Anchor Check, Session Scoping, Drift Trigger -> [references/anti-drift.md](references/anti-drift.md)95- Verification, git hygiene, dependency control, evidence -> [references/operations.md](references/operations.md)96- Live research, anti-default stack choices, product-shape fit -> [references/research-and-fit.md](references/research-and-fit.md)9798## Drift Signals - Hard Stop99100Stop and re-anchor immediately if any of these appear:101102| Signal | Action |103|--------|--------|104| Began implementing without a confirmed Scope Gate | Stop, produce Scope Gate, await confirmation |105| Began implementing without a confirmed Plan Gate | Stop, produce Plan Gate, await confirmation |106| Touched a file not declared in Scope Gate | Stop, declare scope expansion, await approval |107| Guessed library syntax without searching docs | Search live docs, correct output |108| Used stale training data for current facts | Run live research or mark unverified |109| Chose a default stack/product shape without fit check | Compare alternatives against project constraints |110| Unexplained contiguous block > limit | Run `warden-lint.js`, split if needed |111| Skipped Blast Radius Check before a rewrite | Run check before proceeding |112| Claimed completion without verification evidence | Run relevant checks or state residual risk |113| Changed dependencies without version/source evidence | Stop, inspect package metadata and lockfile |114| Edited in a dirty repo without checking ownership | Inspect status and preserve user changes |115| No `[AWAITING CONFIRMATION]` before >2-file change | Pause and request confirmation |116| Monolithic file output without module split | Refactor into separated concerns |117118All limits and thresholds are defined in `codewarden.json`.
aibot88