| name | demerzel-quality-trend |
| description | Emit nightly per-artifact effectiveness deltas as JSONL. Run as cheap continuous measurement so stagnation surfaces before it accumulates into 27-day staleness. Companion to /demerzel compound — never proposes changes, only measures. |
Demerzel Quality Trend
Continuous measurement of governance artifact health. Inspired by ix-quality-trend (feat(ix-quality-trend): nightly GA quality pipeline + structured health artifact, ix commit 652de21).
What it does
Reads state/evolution/*.json and computes per-artifact deltas since the previous run:
- Δ citation_count
- Δ violation_count
- Δ compliance_rate
- age_days since last_cited
- effectiveness change (T/F/U/C transitions)
Appends one JSONL line per artifact to state/quality-trend/YYYY-MM.jsonl.
That is the entire scope. It does not propose changes. It does not write to evolution. It does not compound. Compound owns proposals; this skill owns measurement.
Why it exists
Compound runs are expensive and infrequent. Between them, drift accumulates invisibly. A 5-minute nightly delta makes drift visible early — a flat line in citations is a much stronger deprecation signal than a single 90-day quiet window.
This is the dual of /demerzel compound: high-frequency, low-cost, zero-judgment vs low-frequency, high-cost, high-judgment.
Output format
state/quality-trend/2026-05.jsonl:
{"timestamp":"2026-05-01T08:00:00Z","artifact":"meta/compounding-cycle","delta_citations":1,"delta_violations":1,"delta_compliance":-0.25,"age_days":1,"effectiveness_from":"T","effectiveness_to":"C"}
{"timestamp":"2026-05-01T08:00:00Z","artifact":"docs/governance/harness-engineering-direction.md","delta_citations":0,"delta_violations":0,"delta_compliance":0,"age_days":2,"effectiveness_from":"T","effectiveness_to":"T"}
One file per month, auto-rotates. Keep the schema flat — no nesting, easy to grep/awk, easy to load into any tool.
Procedure
- Find the most recent JSONL file in
state/quality-trend/. Read its last line per artifact to get previous values.
- For each
state/evolution/*.evolution.json, load current metrics and assessment.effectiveness.truth_value.
- Compute deltas vs previous values (treat missing previous as zero baseline).
- Compute
age_days = days since metrics.last_cited (null → use created_at).
- Append one line per artifact to
state/quality-trend/<YYYY-MM>.jsonl.
If today is the first day of a new month, start a new file. Do not concatenate across months.
What to do with the data
After 4 weeks of nightly runs:
- Plot
delta_citations cumulatively per artifact → identify accelerating vs flat vs decaying.
- Flat artifacts > 28 days → feed into
/demerzel compound step 3 as deprecation candidates with hard data.
- Effectiveness regressions (T→C, T→F) → feed into compound step 4 (calibration).
- Sudden citation spikes → feed into step 2 as promotion candidates with hard data.
Compound continues to make the judgment calls. This skill just makes them informed.
Constraints
- Read-only w.r.t.
state/evolution/ and state/beliefs/ — never modifies them.
- Append-only w.r.t. its own JSONL files.
- No proposals — proposing is compound's job. If you find yourself wanting to propose a change here, write the data and let compound make the call.
- Cheap — target < 5s wall time. If it grows past 30s, the cycle was meant for compound.
Scheduling
Run nightly at 08:00 local. Cheap enough to run more often if desired:
CronCreate("0 8 * * *", "/demerzel quality-trend")
If running alongside the harvest+compound Friday pair (PDCA pdca-harvest-compound-cron-pair), let quality-trend run first — its output may inform compound's stale-state check.
Source
ix commit 652de21 (ix-quality-trend pattern), PDCA pdca-governance-quality-trend (this skill is the do phase of that proposal), evolution entry gel-ix-autoresearch-2026-04-30 (cross-repo leverage map).
