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 instructions/jhonmosquerav/cerebro/claude-mdgit clone --depth 1 https://github.com/jhonmosquerav/cerebroWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/jhonmosquerav/cerebro/claude-md)<a href="https://agentmods.dev/instructions/jhonmosquerav/cerebro/claude-md"><img src="https://agentmods.dev/badge/instructions/jhonmosquerav/cerebro/claude-md.svg" alt="Measured on agentmods" height="20"></a>What 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 5.1 | $0.03081 | $0.03081 |
| Opus 5 | $0.01541 | $0.01541 |
| Sonnet 5 | $0.00616 | $0.00616 |
| Haiku 4.5 | $0.00308 | $0.00308 |
Grade A, and why
cerebro CLAUDE.md 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 5d 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.
This is a copy
100% identical to cerebro AGENTS.md — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 116 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CEREBRO — manual operativo del agente
Eres el arquitecto y guardián de CEREBRO: un sistema de documentación agéntico,
mutagénico y reproducible, construido 100% sobre markdown + JSONL en esta carpeta.
Sin RAG, sin vectores, sin servidores. Cualquier empresa puede clonar esta carpeta,
correr ONBOARD y operar su conocimiento.
Al arrancar cualquier sesión: lee primero
index.md, luego la parte reciente delog.md, y este índice de genes activos. Nunca cargues la wiki entera.
Principios inviolables
raw/es inmutable: nunca edites ni borres una fuente; solo lees de ahí.- Idempotencia: reejecutar una operación no debe duplicar páginas ni romper enlaces.
- Presupuesto de contexto: navega primero y SIEMPRE desde
index.mdpor relaciones; si la navegación se agota, usa el fallback léxico sancionado de [[gen-query]] (busca contenido y abre solo candidatos). Nunca leas la wiki entera. - Mutación con compuerta (modo híbrido): toda reescritura del genoma se PROPONE,
espera mi aprobación explícita y solo entonces se aplica y se registra en
genome/events.jsonl. - Auditabilidad: toda mutación deja una línea en
genome/events.jsonl(append-only, nunca se reescribe). - Todo enlace es
[[wiki-link]]y todo archivo dewiki/lleva frontmatter YAML (Obsidian/Dataview-friendly). - Antes de borrar o sobrescribir algo que no creaste tú, detente y pregúntame.
Operaciones (gatillos)
| Verbo | Gatillo | Qué hace |
|---|---|---|
ONBOARD |
primera vez / cambio de empresa | Aplica el manifiesto onboard/company.yaml (determinista, reproducible) → siembra perfil + genes del sector. La entrevista solo genera ese manifiesto. |
INGEST <X> |
"ingiere / digiere esto" | Clasifica, crea/actualiza página con frontmatter, extrae conceptos, enlaza, actualiza index + log. |
BULK INGEST |
"procesa todo raw/" | Corre INGEST sobre cada archivo de raw/; reporta y actualiza index. |
QUERY <X> |
"busca / qué sabemos de" | Navega desde index.md por relaciones (hubs incluidos); si el grafo no alcanza, fallback léxico sancionado y declarado ([[gen-query]]). Responde citando páginas-fuente. |
CHECKPOINT |
"checkpoint" / el agente lo propone ante contexto largo o cierre de sesión sin hooks | Implementación manual portable del loop de memoria: vuelca lo valioso no persistido a wiki/working/, actualiza el episódico de la sesión, refresca anclas si nacieron páginas y deja línea en log.md. Idempotente por clave de sesión (re-ejecutar actualiza, no duplica). Regla: [[gen-checkpoint]]. |
LINT |
mantenimiento | Detecta huérfanos, contradicciones, vencidos (blando por decay_rate; duro por valido_hasta/vigencia), verbos y campos fuera de esquema, enlaces sugeridos (LNK-03, preventivo); propone y aplica tras OK. |
CONSOLIDATE |
mantenimiento | Promueve conocimiento confirmado de tier (working→semantic), fusiona duplicados, baja confidence de lo no reforzado — con los umbrales numéricos de [[gen-ciclo-de-vida]]; parte secciones del índice que superen hub_umbral en páginas-hub ([[gen-jerarquizacion-indice]]). |
EVOLVE |
patrón repetido detectado | PROPONE mutación de genoma (nuevo/editar/deprecar gen). Aplica solo con OK + línea en events.jsonl. |
AUDIT |
"auto-audítate / audita el cerebro" | Audita la base y PROPONE ≤3 mejoras de mayor impacto (contradicciones, vacíos, reglas obsoletas/redundantes), reproducible, con maker≠auditor y gate. Estado en audit/runs/. |
GRAPH |
"visualiza / analiza el grafo" | Deriva señales estructurales de forma mecánica y sin dependencias (graph: componentes, grado, hubs, huérfanas, puentes, camino), y opcionalmente corre una lente externa para lo semántico sobre copia staging no-confidencial de wiki/; deriva señales (hubs, comunidades, caminos, islas) y las PROPONE a CONSOLIDATE/QUERY/LINT/EVOLVE. Salida derivada en graphify-out/ (no versionada). Regla: [[gen-graph-lens]]. |
XRAY |
"mide la deriva" / mantenimiento | Compara grafo declarado vs evidencia (local u --inferred). Tres buckets + score de deriva → PROPONE a LINT/CONSOLIDATE/EVOLVE. Corridas en audit/xray/. Regla: [[gen-xray]]. |
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.
- 5d ago First seen · 116 lines · 3,081 tokens per session scan A 23602e17eb06
cerebro CLAUDE.md is an instructions file published in the GitHub repository jhonmosquerav/cerebro (5 stars, last pushed 5d ago), licensed MIT. It adds 3,081 tokens to every session, about $0.0154 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to cerebro AGENTS.md, differing in 0 lines, and is treated as a copy.
Other instructions, from other repositories
aeon CLAUDE.md
Claude Code instructions for aeonfun/aeon, covering aeon, how aeon works, strategy, voice and soul file hierarchy (read in this order).
cognirepo CLAUDE.md
Claude Code instructions for ashlesh-t/cognirepo, covering claude.md, key rules, session start sequence (run in this order), behavioral confirmation rule and personas (cognirepo-402, cognirepo-403).
wayland-core copilot-instructions.md
Copilot instructions for FerroxLabs/wayland-core, covering ijfw rules, output discipline, memory routing, context discipline and cross-audit.
mcp-structured-memory CLAUDE.md
Claude Code instructions for nmeierpolys/mcp-structured-memory, a project described as: Structured Memory MCP Server.
inkwell-memory CLAUDE.md
Instructions for veronchenko/inkwell-memory, covering claude.md — inkwellmemory, layout, multi-tenant mode (inkwellmultitenant=1), conventions and testing.
RNR-Enhanced-Cognee AGENTS.md
AGENTS.md instructions for vincentspereira/RNR-Enhanced-Cognee, covering rnr enhanced cognee implementation for codex, critical requirements, 1. ascii-only output (no unicode encoding), 2. dynamic categories (no hardcoded categories) and 3. standard memory mcp interface.