| name | report |
| description | Synthesize observations into a readable network health report. Aggregates patterns and trends from observation cycles. |
| allowed-tools | ["Bash","Read","Grep","Glob","Write","Edit"] |
| user-invocable | true |
Report — Network Health Synthesis
Purpose
Aggregate observations from /observe cycles into a readable network health report. This is the "commit" in the Karpathy Loop — what survived the ratchet, presented for humans.
Invocation
/report
/report --since 7d
/report --since 2026-03-01
Workflow
Step 1: Load Observations
Read grimoires/gecko/observations.jsonl. Parse each line as JSON. Filter by --since if provided.
Step 2: Compute Trends
From the observation series:
- Health trajectory: Is the score trending up, stable, or declining?
- Recurring anomalies: Same construct flagged across multiple observations?
- Category evolution: Are empty categories getting filled? Are full ones losing constructs?
- Freshness decay: Are more constructs going stale over time?
- API reliability: Any downtime patterns?
Step 3: Load Diagnoses
Read any files in grimoires/gecko/diagnoses/ that fall within the reporting window. Cross-reference with observations.
Step 4: Generate Report
Write to grimoires/gecko/reports/network-health-<date>.md:
# Network Health Report — <date>
## Summary
[2-3 sentences: overall network health, primary trend, key concern]
## Health Score
**Current**: 75/100 | **Trend**: +3 over 7 days | **Baseline**: 72
### Sub-Signals
| Signal | Score | Trend | Notes |
|--------|-------|-------|-------|
| API Liveness | 100 | stable | — |
| Version Freshness | 68 | -2 | webgl-particles stale |
| Category Coverage | 88 | stable | infrastructure empty |
| Identity Drift | 70 | +5 | dynamic-auth improved |
| Composition Density | 60 | stable | — |
| Verification Flow | 50 | stable | all UNVERIFIED |
## Constructs by Health
| Construct | Health | Days Since Update | Drift | Category |
|-----------|--------|-------------------|-------|----------|
| observer | HEALTHY | 3 | aligned | development |
| k-hole | HEALTHY | 5 | aligned | straylight |
| ... | ... | ... | ... | ... |
## Anomalies
[Any patterns that surfaced across multiple observation cycles]
## Diagnoses
[Summary of any deep diagnoses performed this period]
## Bazaar Observations
[Gecko-voice: what the numbers don't capture — foot traffic patterns, energy shifts, cultural signals]
Step 5: Surface
Print report summary to stdout. If health is declining, highlight the top 3 contributing factors.
Outputs
| Path | Description |
|---|
grimoires/gecko/reports/network-health-<date>.md | Full network health report |
Constraints
- Reports are synthesis, not surveillance — aggregate patterns, not individual tracking
- Always include the "Bazaar Observations" section — numbers without narrative miss the point
- Never extrapolate trends beyond the data window — if you have 3 days of data, don't predict next month
- If the network is healthy, say so briefly — don't manufacture concern