Aicrew dev

Skill AKSoftCode/aicrew/codex-skills/aicrew-dev

TDD-first AI dev pipeline for Claude Code, Cursor, Codex, Gemini CLI & Antigravity. /dev /fix /quick + token-saving hooks. npx aicrew install

Install

npx -y skills add AKSoftCode/aicrew --skill aicrew-dev

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Run the aicrew /dev pipeline in Codex (intake -> research -> design -> TDD -> tests -> security -> audit -> infra -> conclude).

SKILL.md

2.6 KB, as published. Nobody here has run it

aicrew /dev (Codex)

Codex does not support slash commands. Use this skill when the user asks for "/dev" or a full SDLC pipeline.

Token foundation (mandatory)

/dev, /fix, and /quick all share the same 11-capability token-saving stack — only pipeline depth differs. Full reference: ~/Agents/docs/token-foundation.md. Stack: graph-first (codebase-memory-mcp), speculative Scout → verify (SCOUT schema, two-model routing), Karpathy guardrails, layered guardrails (guardrails-taxonomy.md), context-economy read policy, security-guard.py hooks, .ai/state checkpoints, /compact between phases, /handoff on tool switch, optional context-mode + token-optimizer-mcp, caveman default output. For /dev, Scout opens Phase 1 Research before any Glob/Grep/Read; re-scout between phases if context grew.

Default output

Caveman/lean style by default. See ~/Agents/agents/caveman.md and ~/Agents/agents/context-economy.md. /normal or /lean off restores verbose.

Source of truth:

~/Agents/commands/dev.md
Project overrides in .ai/skills/ and repo AGENTS.md (if present)

Token foundation (mandatory — all phases):

Graph-first research: codebase-memory-mcp (search_graph → trace_path → get_code_snippet) before any Grep/Read
Speculative context: Scout pass (cheap model) at start of Phase 1; emit SCOUT: schema; verify before main trace
Layered guardrails: security-guard.py (input) → scope lock (Phase 0) → karpathy-guardrails (Phase 4) → security-reviewer (Phase 6)
Context economy: always on; /compact between phases; /lean amplifies
Two-model routing: Scout on haiku/mini; Research+Implement on sonnet/opus
See: ~/Agents/docs/token-foundation.md

Workflow summary:

Intake: clarify bug/feature/refactor, acceptance criteria, scope, risks, test plan.
Research: Scout pass (graph-first) → verify SCOUT schema → confirm root cause or key code paths.
Brainstorm: 3 alternatives with trade-offs (features/refactors).
Design: interface spec, contract checks, over/under-engineering flags.
Implement: load karpathy-guardrails; TDD first (RED -> GREEN -> REFACTOR per acceptance criterion).
Tests: targeted automated tests + smoke path.
Security: changed files only, no false positives.
Audit: if project defines domain audit checks.
Conclude: summary, tests run, risks, commit message.