body-check — Claude Code skill: what it does, how to install
The body-check Claude Code skill reads existing signals about context, memory, and pending work. It reports the current state and suggests a next action.
MIT ★ 26 updated 2026-11-16
npx skills add wrg32786/aigent-os --skill body-check -a claude-codeWhat body-check does
The skill reads memory files to report context and token pressure, decisions due for review, active priorities, open delegations, and staged memory candidates. It does not keep a separate state; it computes the report when invoked and can cache it in memory/BODY_STATE.json for the session.
The first matching condition determines the suggested response: critical context pressure maps to capsule, a candidate backlog above 10 to digest, and a missing Hestia sweep date or one older than 7 days to sweep. It runs on /open, before /close, when asked for a status check, or before a heavy multi-agent task. Hook-based polling is explicitly ruled out. Unread messages are counted only with --full, which uses the Comms API.
What it helps with
- check context and token pressure
- surface decisions due for review
- count open delegations and staged memory candidates
- choose a suggested action from current signals
Example prompts for the agent
- Run
/body-checkand report my current state. - Check my context and memory pressure.
- How am I doing? Give me a status check.
How to install body-check
1Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill body-check -a claude-code2Manually: put the files in place
~/.claude/skills/body-check/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/body-check/
1Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill body-check -a codex2Manually: put the files in place
.agents/skills/body-check/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill body-check -a cursor2Manually: put the files in place
.cursor/skills/body-check/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install wrg32786/aigent-os skills/body-check2Manually: put the files in place
.github/skills/body-check/SKILL.md1Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill body-check -a gemini-cli2Platform CLI
gemini skills install https://github.com/wrg32786/aigent-os --path skills/body-checkManually: put the files in place
.devin/skills/body-check/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add wrg32786/aigent-os --skill body-check -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about body-check
How do I install body-check in Claude Code?
Run npx skills add wrg32786/aigent-os --skill body-check -a claude-code. The skill file is at ~/.claude/skills/body-check/SKILL.md.
What does body-check do?
It reads existing memory signals and produces a status report with a suggested response. It computes the state when called rather than maintaining a separate one.
Which signals does body-check read?
It reads memory files for heat, usage, decisions, priorities, delegations, candidates, and Hestia sweep history. With --full, it also uses the Comms API to count unread messages.
Source text
SKILL.md · MIT
---
name: body-check
agent: none
description: "Report body state, five pressures, and a recommended reflex from existing signals. Use when checking context pressure, memory pressure, or vital signs."
allowed-tools: Read, Bash
user-invocable: true
triggers:
- body check
- body state
- how am I doing
- check pressure
- context pressure
- memory pressure
- somatic check
- vital signs
---
# /body-check
Reports current body state by reading existing signals — does NOT maintain a separate state. Computed on invocation, optionally cached to `memory/BODY_STATE.json` for the session.
## Sources read
- `memory/HEAT_INDEX.json` → attention/decay state
- `memory/usage_log.md` → token pressure (latest entry vs trailing 5)
- `memory/DECISION_LOG.md` + `DECISION_OUTCOMES.md` → decisions due (30/60/90 day intervals)
- `memory/ACTIVE_PRIORITIES.md` → tier-1 / mode
- `memory/DELEGATION_TRACKER.md` → open delegations
- `memory/MEMORY_CANDIDATES.md` → staged backlog count
- `memory/HESTIA_SWEEP_LOG.md` → last Hestia sweep date (v0.4.4)
- Comms API → unread count (only if invoked with `--full`)
## Output schema
Matches `memory/BODY_STATE.json` `_schema` block exactly. See `system/15_somatic_layer.md` for field definitions and pressure thresholds.
## hestia_last_sweep computation (v0.4.4)
Read `memory/HESTIA_SWEEP_LOG.md`. Find the most recent `### YYYY-MM-DD` entry (matches `^### (\d{4}-\d{2}-\d{2})` regex). The newest matching date becomes `hestia_last_sweep` as ISO8601 (`<date>T00:00:00Z`). If no matches, `hestia_last_sweep = null`.
## Recommended reflex logic
Priority order — first match wins:
1. `context_pressure: critical` → recommend `capsule`
2. `memory_candidate_backlog > 10` → recommend `digest`
3. `hestia_last_sweep` is null OR more than 7 days ago → recommend `sweep` (v0.4.4)
4. `decision_reviews_due > 0` → recommend (no action, surface in /open)
5. All low → `none`
## When to run
- On `/open` (auto, briefly)
- Before `/close` (to inform the close protocol)
- Anytime principal asks "how am I doing"
- Before dispatching a heavy multi-agent task (catches "you're already at high context pressure")
Do NOT run on a hook polling cadence.
© wrg32786 · MIT · wrg32786/aigent-os