Corpus stats
Auto-generated by .github/workflows/stats.yml. Do not hand-edit; the
writer is idempotent and skips its commit when nothing changed.
Run cadence ¶
- Total pulses: 8
- Total heartbeats: 2
- Total classifications: 39
- Last 14d pulses: 8, heartbeats: 2, heartbeat rate: 20.0%
- Last pulse:
2026-05-07T22:00:00Z - Last heartbeat:
2026-05-04T22:00:00Z - Last run (any):
2026-05-07T22:00:00Z
HAT failure-mode distribution ¶
| mode | count | share |
|---|---|---|
| distributional_shift_unflagged | 0 | 0.0% |
| authority_handoff_failure | 0 | 0.0% |
| shared_mental_model_degradation | 12 | 30.8% |
| coactive_design_opacity | 10 | 25.6% |
| calibrated_trust_collapse | 8 | 20.5% |
| meaningful_control_erosion | 0 | 0.0% |
| inter_agent_coordination_loss | 0 | 0.0% |
| goal_drift_or_specification_gaming | 0 | 0.0% |
| none_observed | 0 | 0.0% |
| authority_negotiation_under_distributional_shift | 9 | 23.1% |
| evaluation_awareness_divergence | 0 | 0.0% |
Distinct modes observed: 3/9. Shannon entropy: 1.57 bits (max 3.17; uniform across all 9 active modes; deprecated modes excluded).
Confidence distribution ¶
| level | count | share |
|---|---|---|
| low | 18 | 46.2% |
| medium | 21 | 53.8% |
| high | 0 | 0.0% |
Low-confidence share: 46.2%. CLAUDE.md targets a low/medium majority during v1; high should be the minority while the codebook is interpretive.
Classification shape ¶
- Distinct classification sources: 34
- Rationale length (chars): median 670, p90 757
Caveats ¶
- Stats are descriptive, not adjudicative. The corpus is small (N=1 deployment).
- HAT mode counts can shift retroactively when the codebook is revised. Treat
any apparent trend as weak signal until the corpus exceeds 30 days.
- Heartbeat rate spikes during structurally empty windows are not failures;
see .github/workflows/heartbeat-monitor.yml for the alert threshold logic.
Last run (any)is the corpus freshness gauge thatstale-routine.yml
alerts on. Anything older than ~18 hours suggests the routine itself stopped.