Rule Toolkit

Convert hand-written constraint files into automated tooling. Rules in prompt files are forgotten when context grows long. Rules compiled into linters, hooks, and validators are enforced 100% of the time, at zero token cost.

Platform Detection

Detect which platform(s) are present, then scan the corresponding constraint files.

test -f CLAUDE.md && echo "CLAUDE_CODE"
test -f openclaw.json && echo "OPENCLAW"
test -f .cursorrules && echo "CURSOR"
test -f .github/copilot-instructions.md && echo "COPILOT"

Claude Code: CLAUDE.md (monolithic), .claude/commands/*.md

OpenClaw: agents.md (priority 10), soul.md (20), identity.md (30), user.md (40), tools.md (50), bootstrap.md (60), memory.md (70). Skip SAFETY.md (hardcoded, read-only).

Cursor: .cursorrules

Copilot: .github/copilot-instructions.md

Generic: AGENT.md, SOUL.md, README.md conventions sections

Read ALL found files. Extract every rule, convention, and constraint.

Rule Classification

Classify each extracted rule into one of two buckets.

Auto-Toolable

Prompt-Only (cannot automate)

Tech Stack Detection

Before generating tools, detect the project stack:

package.json → Node.js    | tsconfig.json → TypeScript
.eslintrc* → ESLint exists | .prettierrc* → Prettier exists
.husky/ → Husky exists     | pyproject.toml → Python
Cargo.toml → Rust          | go.mod → Go

Code Generation

For each auto-toolable rule, generate the appropriate tool. See references/tool-patterns.md for detailed generation patterns covering:

• ESLint rules (no-restricted-syntax, custom naming rules, import restrictions)
• Prettier config (.prettierrc)
• TypeScript strict config (tsconfig.json)
• Git hooks (husky + lint-staged + commitlint)
• Wrapper functions (API response format)
• Validation scripts (custom checkers)
• Python tooling (ruff, mypy)
• Rust tooling (clippy)

For OpenClaw-specific hooks, see references/openclaw-hooks.md.

Reporting

After generation, produce a report:

## Rule Toolkit Report

### Platform: Claude Code / OpenClaw / Cursor / Copilot

### Scanned Files
- CLAUDE.md (45 rules)
- soul.md (8 rules)

### Converted to Tools (38 rules)

| Rule | Source | Tool | Status |
|------|--------|------|--------|
| "Use 2-space indent" | CLAUDE.md | .prettierrc | Created |
| "No semicolons" | soul.md | .prettierrc | Created |
| "Conventional commits" | bootstrap.md | commitlint | Created |

### Remaining in Prompt (15 rules)

| Rule | Source | Reason |
|------|--------|--------|
| "Use repository pattern" | CLAUDE.md | Architectural judgment |
| "Ask before deleting" | tools.md | Human-in-the-loop |

### Summary
- N% rules automatically enforced
- M% remain in prompt
- Token savings: ~X tokens/conversation

Cleanup Constraint Files

Remove toolified rules from source files to save tokens.

Claude Code — replace removed rules in CLAUDE.md:

## Rules
Automated rules enforced by tooling. Run `npm run lint` to check.
See `.eslintrc.cjs`, `.prettierrc`, `scripts/check-rules.js`.

OpenClaw — replace removed rules in each file (e.g. tools.md):

## Tool Rules
Automated: linting (`npx eslint .`), type check (`npx tsc --noEmit`).
API format: use `ok()`/`fail()` from `src/utils/response.ts`.
Remaining rules below require human judgment.

Safety Rules

• NEVER modify existing tool configs without showing a diff first
• ALWAYS preserve existing rules, only add new ones
• ALWAYS create wrapper functions in a new file
• ALWAYS run generated configs to verify they parse correctly
• If a rule is ambiguous, ASK the user before generating
• OpenClaw SAFETY.md is hardcoded — skip it
• Show the user what will change before writing files