By enqack
Iterative self-referential agent loop for Claude Code (enqack edition). Bundles agentic development lifecycle skills (start, plan, execute, review, finish) and the CVR audit runtime (activity ledger, run management, lessons-learned). Feeds the same prompt back on exit until a completion promise is detected or max iterations is reached.
Cancel active Enqack Loop
Generate a Conventional Commit message from changes since the last commit
Establish project intent before any planning or execution
Execute an approved implementation plan, collect evidence, and summarize results
Unified finishing sequence: verify → review → commit message
Modifies files
Hook triggers on file write and edit operations
Runs pre-commands
Contains inline bash commands via ! syntax
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A custom Claude Code plugin implementing the Enqack Loop — a self-referential iterative agent loop based on Geoffrey Huntley's "Ralph Wiggum technique."
The Enqack loop is a Bash loop: a
while truethat repeatedly feeds an AI agent a prompt file, allowing it to iteratively improve until done.
This plugin does it inside your current Claude Code session via a Stop hook — no external shell loop required.
You run ONCE:
/enqack-loop "Build a REST API" --completion-promise "DONE" --max-iterations 20
Claude Code automatically:
1. Works on the task
2. Tries to exit
3. Stop hook (hooks/stop-hook.sh) intercepts → {"decision":"block","reason":<prompt>}
4. Same prompt fed back to Claude
5. Claude sees its previous file changes + git history
6. Repeat until <promise>DONE</promise> output OR max iterations reached
State lives in .claude/enqack-loop.local.md (auto-added to .gitignore, never committed).
Each iteration is appended to knowledge-vault/Logs/enqack-loop.log (NDJSON).
| Feature | Official | This Plugin |
|---|---|---|
--dry-run flag | ❌ | ✅ Preview state file without activating |
| Iteration logging | ❌ | ✅ NDJSON → knowledge-vault/Logs/enqack-loop.log |
| Elapsed time tracking | ❌ | ✅ elapsed_seconds in each log entry |
| Auto-.gitignore | ❌ | ✅ State file auto-excluded |
| Project root in system msg | ❌ | ✅ Situational awareness across iterations |
started_at in state | ❌ | ✅ Timestamped start for observability |
Install via the Claude Code plugin marketplace, or manually via the CLI:
claude plugin install enqack-loop
Hooks register automatically — no manual .claude/settings.json configuration needed. Restart Claude Code after installation for hooks to take effect.
/enqack-loopStart an Enqack loop in the current session.
/enqack-loop "PROMPT" [--max-iterations N] [--completion-promise "TEXT"] [--dry-run]
| Option | Default | Description |
|---|---|---|
--max-iterations N | 0 (unlimited) | Stop after N iterations |
--completion-promise TEXT | none | Exact phrase Claude must output to stop |
--dry-run | false | Preview state file without activating |
Examples:
# Run up to 20 iterations, stop when Claude outputs <promise>DONE</promise>
/enqack-loop "Build a REST API for todos. CRUD, validation, tests." \
--completion-promise "DONE" \
--max-iterations 20
# Infinite loop, no completion signal (use /cancel-enqack to stop)
/enqack-loop "Continuously improve the code quality"
# Preview only
/enqack-loop "Fix all lint errors" --max-iterations 10 --dry-run
/cancel-enqackCancel an active Enqack loop (removes the state file).
/cancel-enqack
This plugin bundles a full agentic development lifecycle as slash commands. Each command is a structured protocol document that Claude follows.
/enqack-start
└─ /enqack-establish-intent (if project_intent.md missing)
└─ /enqack-prep-context (if context_manifest.md missing)
└─ /enqack-plan-cycle
├─ /enqack-prep-context
├─ /enqack-verify-agenda
├─ /enqack-plan-execution → approval gate →
├─ /enqack-execute-plan
├─ /enqack-post-verify
└─ /enqack-post-execution-review
/enqack-finish
├─ /enqack-post-verify
├─ /enqack-post-execution-review
├─ /enqack-commit-message
└─ /enqack-markdown-checklist
| Command | Description |
|---|---|
/enqack-start | Entry point — ensures intent + context, hands off to plan-cycle |
/enqack-finish | Closing sequence — verify, review, commit message, format |
/enqack-plan-cycle | Full lifecycle orchestrator |
/enqack-establish-intent | Write knowledge-vault/Intent/project_intent.md |
/enqack-prep-context | Load context + produce context_manifest.md |
/enqack-verify-agenda | Classify all AGENDA.md items before planning |
/enqack-plan-execution | Produce implementation_plan.md + .json |
/enqack-execute-plan | Execute approved plan + run verify_all.sh |
/enqack-post-verify | Reconcile AGENDA.md → post_verify_report.md |
/enqack-post-execution-review | Extract lessons + seal run via close_run.py |
/enqack-commit-message | Propose Conventional Commit messages |
/enqack-markdown-checklist | Format + lint all Markdown files |
/enqack-toggle-maintenance-mode | Toggle agent maintenance mode |
The plugin ships a Python runtime in scripts/cvr/ that powers the agentic development commands.
npx claudepluginhub enqack/cai-plugins-enqack --plugin enqack-loopComprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Continuous self-referential AI loops for interactive iterative development, implementing the Ralph Wiggum technique. Run Claude in a while-true loop with the same prompt until task completion.
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Production-grade engineering skills for AI coding agents — covering the full software development lifecycle from spec to ship.
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.