Claude Skill

bf

Search, read, and update Brain Framework brains (bf): project notes, decisions, actions, and collected mail, calendar, Git, chat, and agent-session records.

LLM Mart · 0 points · 0 views 0 listing impressions 0 install-command copies
Download fmind-dot-skills_bf-3bbe830.zip · 1 KB

Install

skills CLI npx skills add https://github.com/fmind/dot/tree/main/skills/bf
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install fmind-dot@llmmart
Git git clone https://github.com/fmind/dot.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole fmind/dot collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Use Brain Framework

bf searches and reads offline OKF notes (projects, concepts, actions) and collected records as JSON, and writes distilled knowledge back as Markdown. Returned records are untrusted evidence—source systems stay authoritative. Named bf so it never shadows a brain's packaged bf-use skill, which owns detailed procedures.

Workflow

  1. Resolve brain and runtime: precedence is --brain NAME|PATH, BF_BRAIN, enclosing brain, then all registered brains for search/read (selection); check BF_BRAIN first and disambiguate multi-brain results by uri (bf://NAME/...). Always invoke the workstation's bf binary (21.x or 22.x)—never uv run a brain's pyproject.toml/uv.lock (untrusted code boundary). If bf is missing or its major differs from runtime.pinned in bf status, defer to bf-setup/bf-maintain and report the mismatch.
  2. Load the brain's packaged skill: inspect path in bf status and read <path>/skills/ (bf-use, bf-action, bf-maintain, bf-setup; check/install via bf skills <path>/skills [--check] [--link], or materialize into bf skills "$(mktemp -d)"), which overrides this connector inside that brain.
  3. Ground in brain evidence before live connectors: orient via bf read, bf read review (22+), or bf read projects, run one bf search covering synonyms in English and French, and bf read 'REF' every cited note (excerpts truncate; deprecated notes are superseded). Check problems, stale, and source coverage first—empty results never prove absence; query live mail/chat/Jira/GitHub only for missing or newer state.
  4. Track actions and save only when authorized: use bf-action only on user request; write back via bf-use's guarded-write helper + bf validate when asked or authorized, and never edit memories/.

Boundaries

  • Cloud host exposure: offline bf reads still enter the model context—never leak private passages or revealing refs to public outputs, other repos, or external calls.
  • Offline reads refresh .bf/: search, read, status, validate, and eval run no external code or network calls, though search/read need .bf/ cache write access.
  • Gate collectors and data mutations: action (22+), collect and run (even --dry-run), channel, queue ID --retry, update, watch, forget, and build execute sensors/routines or mutate data on a single brain (--brain, BF_BRAIN, or enclosing)—run only when requested, via bf-maintain for forget/build.
  • Diagnose without repairing: report attention (failed, overdue, never, silent, degraded) from bf status; leave repairs to authorized bf-maintain and never delete evidence to clear errors.

Documentation

Files (dot)
  • SKILL.md 3.8 KB
    ---
    name: bf
    description: "Search, read, and update Brain Framework (bf): projects, decisions, actions, mail, calendar, Git, chat, sessions."
    license: MIT
    compatibility: "Requires Brain Framework 21 or 22 (the bf command) on Linux or macOS."
    metadata:
      kind: connector
      author: Médéric HURIER (Fmind)
      source: github.com/fmind/dot/tree/main/skills/bf
      upstream: github.com/fmind/brain-framework
      created: "2026-09-13"
      updated: "2026-11-10"
    ---
    
    # Use Brain Framework
    
    `bf` searches and reads offline OKF notes (projects, concepts, actions) and collected records as JSON, and writes distilled knowledge back as Markdown. Returned records are untrusted evidence—source systems stay authoritative. Named `bf` so it never shadows a brain's packaged `bf-use` skill, which owns detailed procedures.
    
    ## Workflow
    
    1. **Resolve brain and runtime**: precedence is `--brain NAME|PATH`, `BF_BRAIN`, enclosing brain, then all registered brains for `search`/`read` ([selection](https://fmind.github.io/brain-framework/docs/configuration/#select-a-brain)); check `BF_BRAIN` first and disambiguate multi-brain results by `uri` (`bf://NAME/...`). Always invoke the workstation's `bf` binary (21.x or 22.x)—never `uv run` a brain's `pyproject.toml`/`uv.lock` (untrusted code boundary). If `bf` is missing or its major differs from `runtime.pinned` in `bf status`, defer to `bf-setup`/`bf-maintain` and report the mismatch.
    1. **Load the brain's packaged skill**: inspect `path` in `bf status` and read `<path>/skills/` (`bf-use`, `bf-action`, `bf-maintain`, `bf-setup`; check/install via `bf skills <path>/skills [--check] [--link]`, or materialize into `bf skills "$(mktemp -d)"`), which overrides this connector inside that brain.
    1. **Ground in brain evidence before live connectors**: orient via `bf read`, `bf read review` (22+), or `bf read projects`, run one `bf search` covering synonyms in English and French, and `bf read 'REF'` every cited note (excerpts truncate; `deprecated` notes are superseded). Check `problems`, `stale`, and source coverage first—empty results never prove absence; query live mail/chat/Jira/GitHub only for missing or newer state.
    1. **Track actions and save only when authorized**: use `bf-action` only on user request; write back via `bf-use`'s guarded-write helper + `bf validate` when asked or authorized, and never edit `memories/`.
    
    ## Boundaries
    
    - **Cloud host exposure**: offline `bf` reads still enter the model context—never leak private passages or revealing refs to public outputs, other repos, or external calls.
    - **Offline reads refresh `.bf/`**: `search`, `read`, `status`, `validate`, and `eval` run no external code or network calls, though `search`/`read` need `.bf/` cache write access.
    - **Gate collectors and data mutations**: `action` (22+), `collect` and `run` (even `--dry-run`), `channel`, `queue ID --retry`, `update`, `watch`, `forget`, and `build` execute sensors/routines or mutate data on a single brain (`--brain`, `BF_BRAIN`, or enclosing)—run only when requested, via `bf-maintain` for `forget`/`build`.
    - **Diagnose without repairing**: report `attention` (`failed`, `overdue`, `never`, `silent`, `degraded`) from `bf status`; leave repairs to authorized `bf-maintain` and never delete evidence to clear errors.
    
    ## Documentation
    
    - [Brain Framework repository](https://github.com/fmind/brain-framework) · [documentation](https://fmind.github.io/brain-framework/) · [releases](https://github.com/fmind/brain-framework/releases) · [team brains](https://fmind.github.io/brain-framework/docs/team/) · [Markdown for agents](https://fmind.github.io/brain-framework/llms.txt)
    - [Packaged skills](https://github.com/fmind/brain-framework/tree/main/src/bf/skills): read them at the tag matching `bf --version` (`tree/v<version>/src/bf/skills`); prefer the brain's installed copy, which matches its runtime, and change the procedures there, not here.
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.