@hivehub/rulebook 5.4.1 → 5.5.0

package/templates/commands/rulebook-task-list.md

@@ -1,42 +1,42 @@
----
-name: /rulebook-task-list
-id: rulebook-task-list
-category: Rulebook
-description: List all Rulebook tasks (active and optionally archived).
----
-<!-- RULEBOOK:START -->
-**Guardrails**
-- Favor straightforward, minimal implementations first and add complexity only when it is requested or clearly required.
-- Keep changes tightly scoped to the requested outcome.
-- Refer to `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines.
-
-**Steps**
-1. **List Active Tasks**:
-   ```bash
-   rulebook task list
-   ```
-   Shows all active tasks with their status (pending, in-progress, completed, blocked).
-
-2. **List Including Archived**:
-   ```bash
-   rulebook task list --archived
-   ```
-   Shows both active and archived tasks.
-
-3. **Review Task Status**:
-   - **pending**: Task not started
-   - **in-progress**: Task being worked on
-   - **completed**: Task finished (ready for archive)
-   - **blocked**: Task blocked by dependency
-
-4. **Select Task to Work On**:
-   - Choose task with highest priority
-   - Check task status before starting
-   - Verify task is not blocked
-
-**Reference**
-- Use `rulebook task show <task-id>` to view task details
-- Use `rulebook task validate <task-id>` to check task format
-- See `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines
-<!-- RULEBOOK:END -->
-
+---
+name: /rulebook-task-list
+id: rulebook-task-list
+category: Rulebook
+description: List all Rulebook tasks (active and optionally archived).
+---
+<!-- RULEBOOK:START -->
+**Guardrails**
+- Favor straightforward, minimal implementations first and add complexity only when it is requested or clearly required.
+- Keep changes tightly scoped to the requested outcome.
+- Refer to `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines.
+
+**Steps**
+1. **List Active Tasks**:
+   ```bash
+   rulebook task list
+   ```
+   Shows all active tasks with their status (pending, in-progress, completed, blocked).
+
+2. **List Including Archived**:
+   ```bash
+   rulebook task list --archived
+   ```
+   Shows both active and archived tasks.
+
+3. **Review Task Status**:
+   - **pending**: Task not started
+   - **in-progress**: Task being worked on
+   - **completed**: Task finished (ready for archive)
+   - **blocked**: Task blocked by dependency
+
+4. **Select Task to Work On**:
+   - Choose task with highest priority
+   - Check task status before starting
+   - Verify task is not blocked
+
+**Reference**
+- Use `rulebook task show <task-id>` to view task details
+- Use `rulebook task validate <task-id>` to check task format
+- See `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines
+<!-- RULEBOOK:END -->
+

package/templates/commands/rulebook-task-show.md

@@ -1,52 +1,52 @@
----
-name: /rulebook-task-show
-id: rulebook-task-show
-category: Rulebook
-description: Show detailed information about a specific Rulebook task.
----
-<!-- RULEBOOK:START -->
-**Guardrails**
-- Favor straightforward, minimal implementations first and add complexity only when it is requested or clearly required.
-- Keep changes tightly scoped to the requested outcome.
-- Refer to `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines.
-
-**Steps**
-1. **Show Task Details**:
-   ```bash
-   rulebook task show <task-id>
-   ```
-   Displays:
-   - Task ID and title
-   - Status (pending, in-progress, completed, blocked)
-   - Created and updated dates
-   - Archive date (if archived)
-   - Proposal summary
-   - Spec files list
-
-2. **Review Proposal**:
-   - Read `proposal.md` to understand why and what changes
-   - Check impact assessment
-   - Verify breaking changes are documented
-
-3. **Review Tasks Checklist**:
-   - Check `tasks.md` for implementation checklist
-   - Identify completed items (`[x]`)
-   - Identify pending items (`[ ]`)
-   - Follow priority order (most critical first)
-
-4. **Review Spec Deltas**:
-   - Check `specs/*/spec.md` files
-   - Verify format compliance (SHALL/MUST, 4 hashtags for scenarios)
-   - Understand requirements and scenarios
-
-5. **Check Task Status**:
-   - Verify task is not blocked
-   - Check if task is ready for implementation
-   - Confirm all prerequisites are met
-
-**Reference**
-- Use `rulebook task list` to see all tasks
-- Use `rulebook task validate <task-id>` to check task format
-- See `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines
-<!-- RULEBOOK:END -->
-
+---
+name: /rulebook-task-show
+id: rulebook-task-show
+category: Rulebook
+description: Show detailed information about a specific Rulebook task.
+---
+<!-- RULEBOOK:START -->
+**Guardrails**
+- Favor straightforward, minimal implementations first and add complexity only when it is requested or clearly required.
+- Keep changes tightly scoped to the requested outcome.
+- Refer to `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines.
+
+**Steps**
+1. **Show Task Details**:
+   ```bash
+   rulebook task show <task-id>
+   ```
+   Displays:
+   - Task ID and title
+   - Status (pending, in-progress, completed, blocked)
+   - Created and updated dates
+   - Archive date (if archived)
+   - Proposal summary
+   - Spec files list
+
+2. **Review Proposal**:
+   - Read `proposal.md` to understand why and what changes
+   - Check impact assessment
+   - Verify breaking changes are documented
+
+3. **Review Tasks Checklist**:
+   - Check `tasks.md` for implementation checklist
+   - Identify completed items (`[x]`)
+   - Identify pending items (`[ ]`)
+   - Follow priority order (most critical first)
+
+4. **Review Spec Deltas**:
+   - Check `specs/*/spec.md` files
+   - Verify format compliance (SHALL/MUST, 4 hashtags for scenarios)
+   - Understand requirements and scenarios
+
+5. **Check Task Status**:
+   - Verify task is not blocked
+   - Check if task is ready for implementation
+   - Confirm all prerequisites are met
+
+**Reference**
+- Use `rulebook task list` to see all tasks
+- Use `rulebook task validate <task-id>` to check task format
+- See `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines
+<!-- RULEBOOK:END -->
+

package/templates/commands/rulebook-task-validate.md

@@ -1,53 +1,53 @@
----
-name: /rulebook-task-validate
-id: rulebook-task-validate
-category: Rulebook
-description: Validate Rulebook task format against OpenSpec-compatible requirements.
----
-<!-- RULEBOOK:START -->
-**Guardrails**
-- Favor straightforward, minimal implementations first and add complexity only when it is requested or clearly required.
-- Keep changes tightly scoped to the requested outcome.
-- Refer to `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines and format requirements.
-
-**Steps**
-1. **Validate Task Format**:
-   ```bash
-   rulebook task validate <task-id>
-   ```
-   Checks:
-   - Purpose section length (≥20 characters)
-   - Requirement keywords (SHALL/MUST)
-   - Scenario format (4 hashtags, not 3)
-   - Given/When/Then structure
-   - Delta headers format (ADDED/MODIFIED/REMOVED/RENAMED)
-
-2. **Review Validation Results**:
-   - **Errors**: Must be fixed before proceeding
-   - **Warnings**: Should be addressed but not blocking
-
-3. **Fix Format Issues**:
-   - **Purpose too short**: Expand "Why" section to ≥20 characters
-   - **Missing SHALL/MUST**: Add SHALL or MUST keyword to requirements
-   - **Wrong scenario format**: Change `### Scenario:` to `#### Scenario:` (4 hashtags)
-   - **Missing Given/When/Then**: Replace bullet points with Given/When/Then structure
-   - **Wrong delta headers**: Use ADDED/MODIFIED/REMOVED/RENAMED, not "New Requirements"
-
-4. **Re-validate After Fixes**:
-   ```bash
-   rulebook task validate <task-id>
-   ```
-   Ensure all errors are resolved.
-
-5. **Common Format Errors**:
-   - ❌ `### Scenario:` (3 hashtags) → ✅ `#### Scenario:` (4 hashtags)
-   - ❌ "The system provides X" → ✅ "The system SHALL provide X"
-   - ❌ `- WHEN user does X THEN Y` → ✅ `Given X\nWhen Y\nThen Z`
-   - ❌ `## New Requirements` → ✅ `## ADDED Requirements`
-
-**Reference**
-- See `/.rulebook/specs/RULEBOOK.md` for complete format requirements
-- Use Context7 MCP to get official OpenSpec format documentation
-- Use `rulebook task show <task-id>` to view task details
-<!-- RULEBOOK:END -->
-
+---
+name: /rulebook-task-validate
+id: rulebook-task-validate
+category: Rulebook
+description: Validate Rulebook task format against OpenSpec-compatible requirements.
+---
+<!-- RULEBOOK:START -->
+**Guardrails**
+- Favor straightforward, minimal implementations first and add complexity only when it is requested or clearly required.
+- Keep changes tightly scoped to the requested outcome.
+- Refer to `/.rulebook/specs/RULEBOOK.md` for complete task management guidelines and format requirements.
+
+**Steps**
+1. **Validate Task Format**:
+   ```bash
+   rulebook task validate <task-id>
+   ```
+   Checks:
+   - Purpose section length (≥20 characters)
+   - Requirement keywords (SHALL/MUST)
+   - Scenario format (4 hashtags, not 3)
+   - Given/When/Then structure
+   - Delta headers format (ADDED/MODIFIED/REMOVED/RENAMED)
+
+2. **Review Validation Results**:
+   - **Errors**: Must be fixed before proceeding
+   - **Warnings**: Should be addressed but not blocking
+
+3. **Fix Format Issues**:
+   - **Purpose too short**: Expand "Why" section to ≥20 characters
+   - **Missing SHALL/MUST**: Add SHALL or MUST keyword to requirements
+   - **Wrong scenario format**: Change `### Scenario:` to `#### Scenario:` (4 hashtags)
+   - **Missing Given/When/Then**: Replace bullet points with Given/When/Then structure
+   - **Wrong delta headers**: Use ADDED/MODIFIED/REMOVED/RENAMED, not "New Requirements"
+
+4. **Re-validate After Fixes**:
+   ```bash
+   rulebook task validate <task-id>
+   ```
+   Ensure all errors are resolved.
+
+5. **Common Format Errors**:
+   - ❌ `### Scenario:` (3 hashtags) → ✅ `#### Scenario:` (4 hashtags)
+   - ❌ "The system provides X" → ✅ "The system SHALL provide X"
+   - ❌ `- WHEN user does X THEN Y` → ✅ `Given X\nWhen Y\nThen Z`
+   - ❌ `## New Requirements` → ✅ `## ADDED Requirements`
+
+**Reference**
+- See `/.rulebook/specs/RULEBOOK.md` for complete format requirements
+- Use Context7 MCP to get official OpenSpec format documentation
+- Use `rulebook task show <task-id>` to view task details
+<!-- RULEBOOK:END -->
+

package/templates/compact-context/_default.md

@@ -1,23 +1,23 @@
-# Post-compaction cheat sheet
-
-This file is re-injected after every conversation compaction. Keep it
-**short (≤50 lines)** and only include directives that must survive
-compaction. Everything else belongs in `CLAUDE.md`, `AGENTS.md`, or
-`.claude/rules/`.
-
-## Critical reminders
-
-- Read `AGENTS.md` and `AGENTS.override.md` before making changes.
-- Never revert or discard uncommitted work — fix forward.
-- Edit files sequentially, not in parallel (see `.claude/rules/sequential-editing.md`).
-- Run diagnostics (`type-check`, `lint`) before running the full test suite.
-- If a fix fails twice, stop and escalate — do not retry a third time.
-- Prefer MCP tools over shell commands when an equivalent MCP tool exists.
-- Save learnings to `.rulebook/learnings/` at the end of significant work.
-
-## Project-specific reminders
-
-(Edit this file to add reminders unique to your project. Examples:
-build command, primary language toolchain, forbidden directories,
-where canonical documentation lives, critical files that must not
-be touched without review.)
+# Post-compaction cheat sheet
+
+This file is re-injected after every conversation compaction. Keep it
+**short (≤50 lines)** and only include directives that must survive
+compaction. Everything else belongs in `CLAUDE.md`, `AGENTS.md`, or
+`.claude/rules/`.
+
+## Critical reminders
+
+- Read `AGENTS.md` and `AGENTS.override.md` before making changes.
+- Never revert or discard uncommitted work — fix forward.
+- Edit files sequentially, not in parallel (see `.claude/rules/sequential-editing.md`).
+- Run diagnostics (`type-check`, `lint`) before running the full test suite.
+- If a fix fails twice, stop and escalate — do not retry a third time.
+- Prefer MCP tools over shell commands when an equivalent MCP tool exists.
+- Save learnings to `.rulebook/learnings/` at the end of significant work.
+
+## Project-specific reminders
+
+(Edit this file to add reminders unique to your project. Examples:
+build command, primary language toolchain, forbidden directories,
+where canonical documentation lives, critical files that must not
+be touched without review.)

package/templates/compact-context/cpp.md

@@ -1,26 +1,26 @@
-# Post-compaction cheat sheet (C/C++ project)
-
-Re-injected after every compaction. Keep ≤50 lines. Edit freely.
-
-## Critical reminders
-
-- Read `AGENTS.md` and `AGENTS.override.md` before changes.
-- `cmake --build <build>` before running tests.
-- Warnings are errors (`-Wall -Wextra -Werror`).
-- Smart pointers only — no raw `new`/`delete` without justification.
-- Edit files sequentially, not in parallel.
-- If a fix fails twice, stop and escalate.
-
-## Build & test quick reference
-
-- **Configure**: `cmake -S . -B build -DCMAKE_EXPORT_COMPILE_COMMANDS=ON`
-- **Build**: `cmake --build build -j`
-- **Test**: `ctest --test-dir build --output-on-failure`
-- **Format**: `clang-format -i <files>`
-- **Sanitize (CI)**: `-fsanitize=address,undefined`
-
-## Forbidden
-
-- No `using namespace std;` in headers.
-- No undefined behavior shortcuts (strict aliasing, signed overflow).
-- No out-of-source build pollution in `build/` committed.
+# Post-compaction cheat sheet (C/C++ project)
+
+Re-injected after every compaction. Keep ≤50 lines. Edit freely.
+
+## Critical reminders
+
+- Read `AGENTS.md` and `AGENTS.override.md` before changes.
+- `cmake --build <build>` before running tests.
+- Warnings are errors (`-Wall -Wextra -Werror`).
+- Smart pointers only — no raw `new`/`delete` without justification.
+- Edit files sequentially, not in parallel.
+- If a fix fails twice, stop and escalate.
+
+## Build & test quick reference
+
+- **Configure**: `cmake -S . -B build -DCMAKE_EXPORT_COMPILE_COMMANDS=ON`
+- **Build**: `cmake --build build -j`
+- **Test**: `ctest --test-dir build --output-on-failure`
+- **Format**: `clang-format -i <files>`
+- **Sanitize (CI)**: `-fsanitize=address,undefined`
+
+## Forbidden
+
+- No `using namespace std;` in headers.
+- No undefined behavior shortcuts (strict aliasing, signed overflow).
+- No out-of-source build pollution in `build/` committed.

package/templates/compact-context/go.md

@@ -1,26 +1,26 @@
-# Post-compaction cheat sheet (Go project)
-
-Re-injected after every compaction. Keep ≤50 lines. Edit freely.
-
-## Critical reminders
-
-- Read `AGENTS.md` and `AGENTS.override.md` before changes.
-- `go vet ./...` before `go test ./...`.
-- `gofmt -s -w .` on every file you touch.
-- `context.Context` as first parameter for I/O or blocking functions.
-- Edit files sequentially, not in parallel.
-- If a fix fails twice, stop and escalate.
-
-## Build & test quick reference
-
-- **Build**: `go build ./...`
-- **Vet**: `go vet ./...`
-- **Test**: `go test -race ./...`
-- **Format**: `gofmt -s -w .` and `goimports -w .`
-- **Mod tidy**: `go mod tidy` after adding imports
-
-## Forbidden
-
-- No `_ = err` — handle or wrap every error.
-- No vendored `/vendor/` unless project explicitly vendors.
-- No new modules without authorization.
+# Post-compaction cheat sheet (Go project)
+
+Re-injected after every compaction. Keep ≤50 lines. Edit freely.
+
+## Critical reminders
+
+- Read `AGENTS.md` and `AGENTS.override.md` before changes.
+- `go vet ./...` before `go test ./...`.
+- `gofmt -s -w .` on every file you touch.
+- `context.Context` as first parameter for I/O or blocking functions.
+- Edit files sequentially, not in parallel.
+- If a fix fails twice, stop and escalate.
+
+## Build & test quick reference
+
+- **Build**: `go build ./...`
+- **Vet**: `go vet ./...`
+- **Test**: `go test -race ./...`
+- **Format**: `gofmt -s -w .` and `goimports -w .`
+- **Mod tidy**: `go mod tidy` after adding imports
+
+## Forbidden
+
+- No `_ = err` — handle or wrap every error.
+- No vendored `/vendor/` unless project explicitly vendors.
+- No new modules without authorization.

package/templates/compact-context/python.md

@@ -1,26 +1,26 @@
-# Post-compaction cheat sheet (Python project)
-
-Re-injected after every compaction. Keep ≤50 lines. Edit freely.
-
-## Critical reminders
-
-- Read `AGENTS.md` and `AGENTS.override.md` before changes.
-- Type-check (`mypy` / `pyright`) before `pytest`.
-- `ruff check` before committing.
-- Edit files sequentially, not in parallel.
-- Never revert uncommitted work; fix forward.
-- If a fix fails twice, stop and escalate.
-
-## Build & test quick reference
-
-- **Install**: `pip install -e .` or `poetry install` or `uv sync`
-- **Type-check**: `mypy .` or `pyright`
-- **Lint**: `ruff check .`
-- **Format**: `ruff format .` or `black .`
-- **Test**: `pytest`
-
-## Forbidden
-
-- No bare `except:` — catch specific exceptions.
-- No new dependencies without authorization.
-- No `rm -rf` in this repo.
+# Post-compaction cheat sheet (Python project)
+
+Re-injected after every compaction. Keep ≤50 lines. Edit freely.
+
+## Critical reminders
+
+- Read `AGENTS.md` and `AGENTS.override.md` before changes.
+- Type-check (`mypy` / `pyright`) before `pytest`.
+- `ruff check` before committing.
+- Edit files sequentially, not in parallel.
+- Never revert uncommitted work; fix forward.
+- If a fix fails twice, stop and escalate.
+
+## Build & test quick reference
+
+- **Install**: `pip install -e .` or `poetry install` or `uv sync`
+- **Type-check**: `mypy .` or `pyright`
+- **Lint**: `ruff check .`
+- **Format**: `ruff format .` or `black .`
+- **Test**: `pytest`
+
+## Forbidden
+
+- No bare `except:` — catch specific exceptions.
+- No new dependencies without authorization.
+- No `rm -rf` in this repo.

package/templates/compact-context/rust.md

@@ -1,28 +1,28 @@
-# Post-compaction cheat sheet (Rust project)
-
-Re-injected after every compaction. Keep ≤50 lines. Edit freely.
-
-## Critical reminders
-
-- Read `AGENTS.md` and `AGENTS.override.md` before changes.
-- `cargo check` before `cargo test` — check is 5–10x faster.
-- `cargo clippy -- -D warnings` must pass.
-- `cargo fmt` before committing — no style debates.
-- Edit files sequentially, not in parallel.
-- Never revert uncommitted work; fix forward.
-- If a fix fails twice, stop and escalate.
-
-## Build & test quick reference
-
-- **Check**: `cargo check --all-targets`
-- **Build**: `cargo build`
-- **Test**: `cargo test --all-features`
-- **Lint**: `cargo clippy -- -D warnings`
-- **Format**: `cargo fmt`
-- **Run**: `cargo run`
-
-## Forbidden
-
-- No `unwrap()`/`expect()` in non-test code without a safety comment.
-- No `unsafe` without a `// SAFETY:` comment.
-- No new crates without authorization.
+# Post-compaction cheat sheet (Rust project)
+
+Re-injected after every compaction. Keep ≤50 lines. Edit freely.
+
+## Critical reminders
+
+- Read `AGENTS.md` and `AGENTS.override.md` before changes.
+- `cargo check` before `cargo test` — check is 5–10x faster.
+- `cargo clippy -- -D warnings` must pass.
+- `cargo fmt` before committing — no style debates.
+- Edit files sequentially, not in parallel.
+- Never revert uncommitted work; fix forward.
+- If a fix fails twice, stop and escalate.
+
+## Build & test quick reference
+
+- **Check**: `cargo check --all-targets`
+- **Build**: `cargo build`
+- **Test**: `cargo test --all-features`
+- **Lint**: `cargo clippy -- -D warnings`
+- **Format**: `cargo fmt`
+- **Run**: `cargo run`
+
+## Forbidden
+
+- No `unwrap()`/`expect()` in non-test code without a safety comment.
+- No `unsafe` without a `// SAFETY:` comment.
+- No new crates without authorization.

package/templates/compact-context/typescript.md

@@ -1,29 +1,29 @@
-# Post-compaction cheat sheet (TypeScript project)
-
-Re-injected after every compaction. Keep ≤50 lines. Edit freely.
-
-## Critical reminders
-
-- Read `AGENTS.md` and `AGENTS.override.md` before changes.
-- `npm run type-check` before `npm test` — type errors are cheaper.
-- `npm run lint` must pass with zero warnings.
-- Edit files sequentially, not in parallel.
-- Never revert uncommitted work; fix forward.
-- If a fix fails twice, stop and escalate.
-- Use MCP tools (`mcp__rulebook__*`) over shell where available.
-
-## Build & test quick reference
-
-- **Install**: `npm install`
-- **Build**: `npm run build` (TypeScript → `dist/`)
-- **Type-check**: `npm run type-check`
-- **Lint**: `npm run lint`
-- **Test**: `npm test` (or `npx vitest run <file>` for single file)
-- **Dev**: `npm run dev`
-
-## Forbidden
-
-- No `any` without justification.
-- No `// @ts-ignore` without a one-line reason.
-- No new dependencies without explicit authorization.
-- No `rm -rf` in this repo.
+# Post-compaction cheat sheet (TypeScript project)
+
+Re-injected after every compaction. Keep ≤50 lines. Edit freely.
+
+## Critical reminders
+
+- Read `AGENTS.md` and `AGENTS.override.md` before changes.
+- `npm run type-check` before `npm test` — type errors are cheaper.
+- `npm run lint` must pass with zero warnings.
+- Edit files sequentially, not in parallel.
+- Never revert uncommitted work; fix forward.
+- If a fix fails twice, stop and escalate.
+- Use MCP tools (`mcp__rulebook__*`) over shell where available.
+
+## Build & test quick reference
+
+- **Install**: `npm install`
+- **Build**: `npm run build` (TypeScript → `dist/`)
+- **Type-check**: `npm run type-check`
+- **Lint**: `npm run lint`
+- **Test**: `npm test` (or `npx vitest run <file>` for single file)
+- **Dev**: `npm run dev`
+
+## Forbidden
+
+- No `any` without justification.
+- No `// @ts-ignore` without a one-line reason.
+- No new dependencies without explicit authorization.
+- No `rm -rf` in this repo.

package/templates/core/AGENTS_LEAN.md

@@ -16,6 +16,15 @@ These override everything else. Violation = output rejected.
 7. **Follow task sequence.** Execute `tasks.md` items in the EXACT order listed. No reordering, no cherry-picking, no starting Phase N+1 before Phase N is 100% done. The list is an ORDER, not a MENU.
 8. **Execute the full task in one turn.** Never stop mid-task to ask "should I proceed?" or "want me to also...?". Make autonomous decisions within scope. Only ask for genuine ambiguity, destructive ops, or impossible tasks.
 
+## Editing Discipline (Karpathy-inspired)
+
+Behavioral guidelines that reduce common LLM coding mistakes. Adapted from [forrestchang/andrej-karpathy-skills](https://github.com/forrestchang/andrej-karpathy-skills), grounded in [Andrej Karpathy's observations](https://x.com/karpathy/status/2015883857489522876).
+
+1. **Think before coding.** State assumptions explicitly. If multiple interpretations exist, present them — don't pick silently. If a simpler approach exists, say so. If something is unclear, stop and ask. Don't hide confusion.
+2. **Simplicity first.** Minimum code that solves the problem. No features beyond what was asked, no abstractions for single-use code, no "flexibility" that wasn't requested, no error handling for impossible scenarios. If you write 200 lines and 50 would do, rewrite.
+3. **Surgical changes.** Touch only what you must. Don't "improve" adjacent code, comments, or formatting. Don't refactor things that aren't broken. Match existing style. If you notice unrelated dead code, mention it — don't delete it. Every changed line must trace directly to the user's request.
+4. **Goal-driven execution.** Define verifiable success criteria upfront. "Add validation" → "write tests for invalid inputs, then make them pass." For multi-step tasks, state a brief plan: `[step] → verify: [check]`. Strong criteria let you loop independently; weak criteria require constant clarification.
+
 ## Critical Rules
 
 - **ALWAYS read `/.rulebook/specs/RULEBOOK.md`** before creating tasks.