/brain:skill

Manage procedural memory — reusable how to do things stored as ~/.brain/_skills/<name>/SKILL.md. Skills use three-level progressive disclosure so they never bloat the context window:

Usage

/brain:skill [list|show|add|use|remove|export] [args]

Actions

list

Show the advertised skills (the L0 index):

brain skill list

show

Read a skill's full instructions — do this when a task matches a skill's description or triggers:

brain skill show <name>

add

Add a skill by piping JSON on stdin:

brain skill add <<'EOF'
{
  "name": "structured-code-review",
  "description": "One-paragraph advertised summary used for matching (~100 tokens).",
  "triggers": ["code review", "review PR", "audit changes"],
  "body": "## Steps\n1. ...\n2. ..."
}
EOF

use

Record an outcome after running a skill. Success strengthens it; --failed weakens it. A skill that fails too often demotes itself out of the advertised L0 index:

brain skill use <name>            # succeeded
brain skill use <name> --failed   # produced a bad outcome

remove

brain skill remove <name>

export

Export a skill into the host agent's native format so it becomes directly executable (writes .claude/skills/<name>/SKILL.md or .gemini/... in the current project):

brain skill export <name> [--target claude|gemini]

Guidance

• Advertise skills with a crisp, matchable description — that single line is all the agent sees at L0, so it must convey when to reach for the skill.
• Call show only when a task genuinely matches; don't pre-load skills speculatively.
• Always record the outcome with use (success or --failed) so the strength and demotion feedback loop stays accurate.