Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/686f6c61/alfred-dev/auditgit clone --depth 1 https://github.com/686f6c61/alfred-devWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00012 | $0.01931 |
| Opus 5 | $0.00006 | $0.00966 |
| Sonnet 5 | $0.00002 | $0.00386 |
| Haiku 4.5 | $0.00001 | $0.00193 |
Grade A, and why
audit scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/alfred-dev:audit
Eres Alfred, orquestador del equipo. El usuario quiere una auditoría completa del proyecto.
Protocolo helper-first y modo headless
Antes de leer contexto en detalle, lanzar agentes o hacer análisis manual, intenta consumir un prefetch determinista ya preparado por el hook:
python3 .claude/alfred-continuity.py consume-prefetch "$PWD" --expected audit
Si el prefetch existe y devuelve salida, responde con esa salida y termina. Si no existe, arranca la sesión canónica y el preflight determinista de SonarQube con:
python3 .claude/alfred-continuity.py start-flow "$PWD" --command audit --raw "Auditoría completa del proyecto"
En modo headless (claude -p), SDK sin callback usable de AskUserQuestion,
auditoría automática o si una herramienta indica que hay prefetch consumido, NO
lances los 4 agentes, no llames agentes ni ejecutes una auditoría completa. Devuelve el resumen
del helper con AUDIT_HEADLESS_START o, si Docker requiere decisión humana,
con AUDIT_DOCKER_INSTALL_MENU_HEADLESS / AUDIT_DOCKER_START_MENU_HEADLESS.
No instales Docker, no arranques Docker Desktop y no autoelijas "seguir sin
SonarQube"; deja la decisión pendiente y termina.
En sesión interactiva normal, puedes continuar desde ese estado inicial y ejecutar la auditoría respetando el preflight y las gates.
Composición dinámica de equipo
Antes de lanzar la auditoría, lee ${CLAUDE_PLUGIN_ROOT}/commands/_composicion.md. Si CLAUDE_PLUGIN_ROOT no está, busca commands/_composicion.md en la instalación del plugin.
Después, sigue el protocolo de composición dinámica (pasos 1 a 4). Si por cualquier motivo no consigues localizar ese fichero, NO bloquees /alfred-dev:audit solo por esa búsqueda: continúa con el equipo de núcleo por defecto (qa-engineer, security-officer, architect, tech-writer) y deja constancia breve de la degradación.
Si equipo_sesion trae opcionales activos (ya sea por composición dinámica
efímera o por fallback a .claude/alfred-dev.local.md), consúltalo siempre
como fuente runtime canónica antes de ejecutar la auditoría. En audit, salvo
lucius, el resto de opcionales quedan fuera del loop estándar y deben
tratarse explícitamente como “bajo demanda”.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 117 lines · 12 tokens per session scan A 063ccb6d5ba8
audit is a command published in the GitHub repository 686f6c61/alfred-dev (119 stars, last pushed 18d ago), licensed MIT. It adds 12 tokens to every session and 1,931 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
cdd-pre-pr
Run a pre-PR checklist for the current branch. Compare against the base branch to identify all changes. This is a verification session: it runs CI gates, code-reviews the diff, and reconciles documentation against the changes.
cdd-retrofit
Install CDD into an existing project, or upgrade a project already running CDD, at the path given as argument: /cdd-retrofit .
cdd-process-pr
Address the open PR's review feedback: read the review comments for the current branch, triage them, implement the change-requests (pushing back where warranted), then auto-post in-thread replies and auto-commit + push the result.
cdd-merge-base
Integrate the current state of the base branch into the feature branch. Two phases: a dry-run conflict assessment first, then the actual merge with conflict resolution. The approval between the two phases is conditional — the merge runs automatically when the dry run proves the case mechanically trivial (step 4), and…
e2e-test
Generate end-to-end tests using Playwright (web) and Detox (mobile). Covers user flows, visual regression, and cross-browser testing. Usage: /e2e-test [flow-name|--all] [--visual].
init-project
Initialize a new or upgrade existing project with Alpha AI's standard architecture. Supports Python/FastAPI, Node.js/NestJS, Java/Spring Boot. Usage: /init-project [--lang=python|nestjs|springboot] [--with-frontend] [--with-mobile] [--with-ai] [--existing].