菜单
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.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| 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. |
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).
Reads state/evolution/*.json and computes per-artifact deltas since the previous run:
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.
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.
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.
state/quality-trend/. Read its last line per artifact to get previous values.state/evolution/*.evolution.json, load current metrics and assessment.effectiveness.truth_value.age_days = days since metrics.last_cited (null → use created_at).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.
After 4 weeks of nightly runs:
delta_citations cumulatively per artifact → identify accelerating vs flat vs decaying./demerzel compound step 3 as deprecation candidates with hard data.Compound continues to make the judgment calls. This skill just makes them informed.
state/evolution/ and state/beliefs/ — never modifies them.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.
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).
Use when a pull request needs review by a different AI model than the one that authored it, or when configuring cross-model code review pipelines in CI/CD.
Run Demerzel's autonomous driver — full cycle or individual phases across all repos
Capture meaningful session state (current cursor, in-flight work, live hypotheses, open questions, do-NOT-carry-forward, success criteria) to state/digests/latest.md so the next session — including one after auto-compaction — can re-enter without re-discovering context cold. Distinct from /learnings (which captures surprises). Validates against docs/contracts/digest-schema.json.
Autonomous research scheduler — selects the most under-researched department by coverage ratio, generates questions from completeness gaps, cross-model validates, auto-produces courses at T>=0.8, rate-limited to 6 cycles/day with kill switch
Personalized AI tutor — turns any topic into a customized, progress-tracked learning course. Use when the user wants to LEARN a concept, skill, language, tool, or domain. Creates a MISSION, curated RESOURCES, iterative lessons with self-checks, and learning-records that persist across sessions. NOT for interrogating the user about their own design (use brainstorming/IDSD for that).
Run one bounded supervised autonomous cycle on the Demerzel governance repo. Picks the smallest unchecked governance slice inside allow_edit, implements it, runs the oracle (`pwsh scripts/verify.ps1`), emits cycle evidence, and stops. Refuses to run unless dev-process-overseer reports loop-eligible and the preflight passes. Use when asked to "run one supervised loop", "advance governance autonomously", or "do one cycle and stop".