body-check

Compose the AIgent's body state by reading existing signals. Lazy compute. Never polled. Reports the five pressures + recommended reflex.

SKILL.md

GitHub

/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.