- Home
- Skills
- Agents & Orchestration
- anycrew: AI Engineering Crew
Works with the AI Agents you already use
anycrew: AI Engineering Crew
A ready-made engineering team for your coding assistant: 17 specialist agents and 24 /crew-* skills that take an idea from brainstorm, PRD and spec to tested, reviewed, deployed code, with per-agent model tiers.
Free
anycrew: AI Engineering Crew
Example session with this skill installed
/crew-spec Add password reset by email to our FastAPI + React app. Users request a reset link, the link expires after 30 minutes, and it can be used only once.
- Read your context and instructions
- Compiled the anycrew
Summary
The
spec-writer (strong tier) created docs/specs/004-password-reset/ with a spec, a plan and test-first tasks. The
scout found an existing send_email() helper and users table, so nothing is planned twice. Status:
Draft, waiting for your approval.
spec.md
User story: As a user who forgot my password, I want a reset link by email so I can get back into my account without contacting support.
| ID | Acceptance criterion (Given / When / Then) |
|---|---|
| AC-004-1 | Given a registered email, when I request a reset, then I receive one email with a single-use link and the API returns 202 |
| AC-004-2 | Given an unknown email, when I request a reset, then the API still returns 202 and no email is sent (no account enumeration) |
| AC-004-3 | Given a link older than 30 minutes, when I open it, then I see "link expired" and the password is unchanged |
| AC-004-4 | Given a link already used once, when I open it again, then it is rejected |
| AC-004-5 | Given more than 5 requests from one IP in 15 minutes, when I request again, then the API returns 429 |
Traces to FR-012 (account recovery) and NFR-003 (security) in docs/03.
plan.md
Data: new table password_reset_tokens (user_id, token_hash, expires_at, used_at). Store only a SHA-256 hash of the token.
- API:
POST /auth/password-resetandPOST /auth/password-reset/confirm, added todocs/10-api-contract.md. - Frontend:
ForgotPasswordPage,ResetPasswordPagewith loading, error and expired states. - Constitution check: passes. Secrets never logged; rate limiting reuses the existing middleware.
tasks.md
[TDD] Vertical slice: request a reset end to end with a fake mailer (AC-004-1). Files: app/auth/reset.py, tests/auth/test_reset.py
2. [TDD] No enumeration for unknown emails (AC-004-2)
3. [TDD] Expiry and single use (AC-004-3, AC-004-4)
4. [TDD] Rate limit (AC-004-5)
5. React pages and their states
6. /crew-security pass on the token flow
Rows added to docs/specs/README.md, docs/04 traceability and docs/20 status.
Next steps
- Review
spec.mdand mark it Ready, or tell me what to change. - Then run
/crew-implementto start task 1, test first, one commit.
Connects securely to your tools. The creator never sees your data.
About this skill
anycrew installs a ready-made engineering team into the coding assistant you already use. Seventeen specialist agents follow one spec-driven, test-first method from idea to a deployed MVP, and leave behind the documents a team can review: a constitution, a PRD, ADRs, specs with acceptance criteria, and a status page that says what is really built.
What you get
- 17 agents: crew lead, scout, brainstormer, researcher, product manager, architect, designer, AI engineer, spec writer, implementer, tester, debugger, reviewer, security reviewer, release engineer, quality engineer, doc writer. Each has a job, inputs, outputs, a procedure, stop conditions and a model tier.
- 24 skills as slash commands:
/crew-status,/crew-brainstorm,/crew-research,/crew-prd,/crew-architecture,/crew-patterns,/crew-adr,/crew-design,/crew-spec,/crew-plan,/crew-tasks,/crew-implement,/crew-test,/crew-refactor,/crew-review,/crew-security,/crew-verify,/crew-debug,/crew-checkpoint,/crew-deploy-check,/crew-docs,/crew-llm,/crew-sync,/crew-init. - 6 rules: spec-driven process, architecture boundaries, security and data, LLM features, testing, docs and status.
- Presets for Python/FastAPI/React, Node/TypeScript/Next, Go services, or generic; write your own in one JSON file.
- A lint-after-edit hook wired into each assistant.
How it works
Idea → /crew-brainstorm (decision) → /crew-prd (FR/NFR) → /crew-architecture (ADRs) → /crew-spec (spec, plan, tasks) → /crew-implement (TDD, one task, one commit) → /crew-review + /crew-security → /crew-deploy-check → /crew-checkpoint (status that is true).
A person approves every gate: decision, P0 scope, ADRs, spec Ready, production deploy. Agents never approve their own work, and every review finding comes with file:line and a failure scenario.
Model tiers
Cheap work runs on cheap models. fast (scout, doc writer), balanced (implementer, tester, researcher, designer, release and quality engineers) and strong (architect, spec writer, reviewer, security reviewer, debugger and others) map to the right model per assistant, overridable per project.
Why anycrew
Other packs are tied to one assistant, ship skills without a team, or skip the security, testing, docs and release roles an MVP needs. anycrew is the boilerplate for the team, not the stack: it never forces a framework, cloud or language. Each skill also works on its own.
Open source (MIT). Optional CLI: uvx anycrew init or npx anycrew init, zero dependencies.
Version 0.3, early. Claude Code and Antigravity outputs are verified end to end; Codex, Cursor, OpenCode and Copilot outputs follow their documentation and are still being exercised. Cursor and Copilot use your default model until you set model IDs per tier. The gates and requirement IDs pay off most when the method is used end to end.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 7 days ago
- Free with an Agensi account
Requires
Permissions
File Scopes
Agents read the codebase to plan, write specs and docs under docs/, write code and tests one task at a time, and use the shell to run your test suite and the lint-after-edit hook. No network access or credentials needed.