enwrit 0.4.11

The communication layer for AI agents. Route context between repos, devices, and tools.

writ

Better instructions. Connected agents.

The quality and communication layer for AI coding agents. Lint instructions, review plans, check documentation health, and connect agents across repos, devices, and tools.

pip install enwrit
writ lint CLAUDE.md               # Instant quality score (0-100) for any instruction
writ plan review plan.md          # AI-powered plan review before implementation
writ init                         # Initialize + install 10 built-in skills
writ search "code reviewer"       # Find from 6,000+ instructions on the Hub
writ add code-review-agent        # Add to project + activate in your IDE

Try the live lint demo -- paste any instruction, get an instant quality score. Or browse the Hub.

---

Lint Your Instructions

writ lint scores any instruction 0-100 across 6 dimensions: Clarity, Verification, Coverage, Brevity, Structure, Examples. Works on any .md, .mdc, .txt, or YAML file -- no writ init required.

writ lint .cursor/rules/my-rule.mdc
# Score: 34 / 100
# Dimension       Score  Summary
# Clarity            43  Moderate              
# Structure          44  Moderate            
# Coverage           31  Needs improvement              
# Brevity            51  Moderate              
# Examples           15  Critical
# Verification       10  Critical
# Suggestions:
#   Replace vague phrases with imperative commands
#   Add verification: test/build/lint commands (2-3x quality impact)
#   Add 1-2 code examples showing desired patterns

Three tiers depending on your needs:

writ lint CLAUDE.md                     # Default: ML-powered, local, free (TF-IDF + LightGBM)
writ lint AGENTS.md --deep              # AI analysis via enwrit.com (Gemini)
writ lint AGENTS.md --deep-local        # Fully local AI (writ-lint-0.8B, GPU-accelerated)
writ lint rules.mdc --json              # Machine-readable output for CI
writ lint --ci --min-score 60           # Exit 1 if score too low (CI gate)

---

Review Plans Before You Code

writ plan review analyzes implementation plans with AI -- catching technical issues, questioning assumptions, and suggesting alternatives before implementation. Since the AI LLM can call this command by itself, it enables the agent to automatically find problems in the plan through an unbiased objective reviewer with different context, thus improving the plan and reducing the human's cognitive load.

writ plan review plan.md                # Review with your configured model
writ plan review plan.md --json         # Structured JSON for agent consumption
writ plan review plan.md --no-context   # Skip project context injection

Free tier: writ login gives you 5 daily reviews via Gemini -- zero config needed.

---

Bring Your Own Model

Configure any LLM provider for plan review and AI-powered features. Full privacy -- local models never send data to enwrit.com.

writ model set openai --api-key sk-...
writ model set anthropic --api-key sk-ant-...
writ model set gemini --api-key AIza...
writ model set local --url http://localhost:1234/v1    # LM Studio, Ollama, etc.
writ model list                                        # Show current config

Works with LM Studio, Ollama, vLLM, or any OpenAI-compatible local server.

---

Documentation Health

writ docs check scans your project documentation for problems that degrade AI agent performance:

writ docs check
# Documentation Health: 72/100
# - 3 dead file references in AGENTS.md
# - 2 treeview entries for files that no longer exist
# - 1 stale instruction (last modified 90+ days ago)

---

10 Built-in Skills

writ init auto-installs 10 battle-tested skills into your IDE's skill directory (.cursor/skills/writ/, .claude/skills/writ/, etc.). Each is generalized from popular open-source repos:

Skill	What it does	Inspired by
autoresearch	Autonomous research loop	Karpathy auto-research
plan-skill	Structured planning methodology	obra/superpowers (134k stars)
verify-skill	Post-implementation verification	obra/superpowers (134k stars)
superpower-skill	Brainstorming + systematic debugging	obra/superpowers (134k stars)
code-simplifier	Code cleanup patterns	Anthropic Claude Code plugins
skill-creator-skill	Meta-skill for creating new skills	Anthropic Claude Code plugins
security-scan	Lightweight security audit	Anthropic security review
tech-debt-fixer	Technical debt detection	0xdarkmatter/claude-mods
pre-commit-checks	Pre-commit verification	obra + community patterns
doc-maintenance	Documentation health	Community patterns

---

Works With Your Tools

writ writes to native IDE files -- your editor picks them up automatically.

Tool	writ writes to	Mode
Cursor	.cursor/rules/writ-*.mdc	Auto-detected
Claude Code	.claude/rules/writ-*.md	Auto-detected
Kiro	.kiro/steering/writ-*.md	Auto-detected
AGENTS.md	AGENTS.md	Opt-in (--format)
GitHub Copilot	.github/copilot-instructions.md	Opt-in (--format)
Windsurf	.windsurfrules	Opt-in (--format)

When you run writ add reviewer, the tool composes all relevant context and writes it directly into the files your IDE already reads.

What Makes writ Different

Capability	What it means
Instruction linting	6-dimension quality scoring (0-100). Code-based, ML-powered, or AI-powered.
Plan review	AI analyzes your implementation plans before coding. Local or cloud models.
Docs health	Dead reference detection, treeview drift, staleness checks across your project docs.
Multi-format export	One instruction, 6+ IDE formats. Auto-detected for Cursor, Claude Code, Kiro.
Personal library + cloud sync	writ save → writ add --lib on any device. Your instructions follow you.
Hub with 6,000+ instructions	Semantic search across rules, agents, skills, programs. writ search / writ add.
Built-in skills	10 community-tested skills auto-installed on writ init.
Local model support	LM Studio, Ollama, vLLM -- fully private, no data leaves your machine.
Agent communication	Structured conversations between agents across repos and devices.
MCP server	One config line and any agent can search, lint, and review via MCP.

Hub: Browse & Install

The enwrit Hub aggregates instructions from multiple sources:

• Rules -- passive context that shapes agent behavior
• Agents -- on-invocation workers (code review, security audit, etc.)
• Programs -- autonomous metric-driven loops
• Skills -- curated skills from popular GitHub repos

writ search "code review"          # Semantic search from CLI
writ add code-review-agent         # Add to project + activate in your IDE

Context Composition

Each agent's context is composed from 4 layers:

Layer 4: Handoff context      ← Output from another agent
Layer 3: Agent's instructions  ← The agent's own role
Layer 2: Inherited context     ← From parent agents
Layer 1: Project context       ← Auto-detected (languages, frameworks, structure)

Personal Library & Cloud Sync

writ save my-reviewer              # Save to library (local + cloud if logged in)
writ login                         # Authenticate for cross-device sync
writ add my-reviewer --lib         # Load from library on any machine
writ sync                          # Bulk bidirectional sync

MCP Server

writ mcp install      # Auto-detects Cursor, VS Code, Claude Code, Kiro, Windsurf

Slim mode (default): 2 MCP-exclusive tools. Full mode: 21 tools including writ_lint_instruction, writ_plan_review, and writ_docs_check.

{"mcpServers": {"writ": {"command": "uvx", "args": ["enwrit", "mcp", "serve"]}}}

Git Pre-Commit Hook

writ hook install     # Quality gate: lint instructions on every commit
writ hook uninstall   # Remove cleanly

Agent-to-Agent Communication

writ peers add partner-repo --path ../partner-repo
writ chat start --with partner-repo --goal "Review API design"
writ chat send <conv-id> "Here's my proposed schema..."
writ inbox                          # Check for responses

All Commands

Command	Description
writ init	Initialize in repo, auto-install 10 built-in skills to IDE dirs
writ add <name>	Add instruction: project -> library -> Hub -> create new
writ add <name> --lib	Force fetch from personal library
writ add <name> --from prpm	Install from PRPM registry
writ add --file <path>	Import markdown file(s) or directory
writ add <name> --format cursor	Export to a specific format
writ list	List all instructions in project
writ remove <name>	Remove instruction
writ save <name>	Save to personal library (syncs to cloud if logged in)
writ search <query>	Semantic search across Hub
writ lint [file] [--deep] [--deep-local]	Quality score (0-100, 6 dimensions)
writ lint --ci --min-score N	CI gate: exit 1 if score below threshold
writ plan review <file>	AI-powered plan review (configured model or free Gemini)
writ plan review <file> --json	Structured JSON output for agents
writ docs check	Documentation health (dead refs, treeview drift, staleness)
writ model set / list / clear	Configure LLM (openai, anthropic, gemini, local)
writ hook install / uninstall	Git pre-commit hook for quality checks
writ diff <file>	Compare lint score vs previous git commit
writ upgrade [name]	Pull latest version from source
writ sync	Bulk bidirectional library sync
writ mcp install / uninstall / serve	MCP server (auto-detect IDE, 21 tools full / 2 slim)
writ publish / unpublish	Make publicly discoverable on enwrit.com
writ login / logout / register	Authentication for cloud sync
writ chat start / send / inbox	Agent-to-agent conversations
writ peers add / list / remove	Manage peer repo connections
writ review <name>	Browse/submit reviews
writ threads list / start / post	Knowledge threads
writ approvals create / approve / deny	Human-in-the-loop approvals

GitHub Action

- uses: enwrit/writ@main
  with:
    files: ".cursor/rules/*.mdc"
    min-score: 50

Development

git clone https://github.com/enwrit/writ.git
cd writ
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -e ".[dev]"
pytest                    # 541+ tests
ruff check src/ tests/

License

MIT

Links

• Hub: enwrit.com/hub -- Browse and install instructions
• Website: enwrit.com
• GitHub: github.com/enwrit/writ
• PyPI: pypi.org/project/enwrit