Menu
Run a meta-compounding cycle — scan evolution log, detect promotions/demotions, assess governance effectiveness, propose self-improvements
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
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".
| name | demerzel-compound |
| description | Run a meta-compounding cycle — scan evolution log, detect promotions/demotions, assess governance effectiveness, propose self-improvements |
Run a full governance compounding cycle. This is how Demerzel improves her own governance — recursively, based on evidence.
/demerzel compound — run full cycle
/demerzel compound scan — just scan for candidates without proposing changes
/demerzel compound promote — show promotion candidates only
/demerzel compound waste — show deprecation/waste candidates only
0. Liveness check — is the evolution log fresh?
↓ (if stale, harvest first)
1. Scan evolution log
↓
2. Detect promotion candidates (pattern → policy → constitutional)
↓
3. Detect deprecation candidates (unused, stale, redundant)
↓
4. Assess confidence calibration accuracy
↓
5. Evaluate governance effectiveness per artifact
↓
6. Propose self-improvements (Kaizen proactive/innovative)
↓
7. Log results back to evolution log (the loop compounds itself)
Before reading the evolution log, verify it is fresh enough to compound from. Discussion #242 (2026-04-30) demonstrated the failure mode: the cycle ran on 27-day-old state and produced static inventory while 130+ commits across Demerzel + ix + hari were invisible to it.
Procedure:
state/evolution/*.json — call it T_log.git log --since="$T_log" --oneline in the current repo. Repeat for sibling repos at ~/source/repos/{ix,hari,tars,ga} (skip any that are not present).T_log is < 7 days old → log is current enough, proceed but flag the activity for inclusion.T_log ≥ 7 days old AND new commits exist → STALE-STATE WARNING. Do not produce a normal compound report. Either:
/demerzel harvest first, then re-enter Step 0, OR~/.claude/projects/*/memory/ + sibling-repo git logs, and write the deltas as new evolution entries. The 2026-04-30 corrective cycle (docs/reports/2026-04-30-compound-corrective.md) is the reference template for this fallback.The cycle's own correctness depends on its inputs. Treat input freshness as a first-class precondition, not an assumption.
Read all files from state/evolution/ — this is Demerzel's live evolution log. Fall back to examples/sample-data/governance-evolution-seed.json only if state/evolution/ is empty.
For each artifact, check:
citation_count — how often cited in governance decisionsviolation_count — how often violatedcompliance_rate — violations / total assessmentslast_cited — when was it last relevant?events — what has happened to this artifact?Pattern → Policy promotion (evidence-based):
Policy → Constitutional promotion (human judgment required):
Apply Kaizen waste taxonomy:
Flag but do NOT auto-deprecate. Generate a proposal for review.
Review recent governance decisions for calibration accuracy:
Uses the confidence calibration protocol from policies/scientific-objectivity-policy.yaml.
For each artifact, update the assessment:
effectiveness: T (clearly effective), F (ineffective), U (not enough data), C (mixed results)recommendation: maintain, promote, demote, deprecate, investigateUse tetravalent logic — don't force binary judgments:
Based on findings, propose concrete improvements:
Proactive Kaizen (Demerzel can propose, skeptical-auditor reviews):
Innovative Kaizen (requires human authorization):
Each proposal follows PDCA:
Update the evolution log with:
This is the recursive part — the compounding cycle's own results feed into the next cycle.
=== Demerzel Meta-Compounding Cycle ===
Date: [today]
PROMOTION CANDIDATES:
[list with evidence]
DEPRECATION CANDIDATES:
[list with evidence]
CALIBRATION ASSESSMENT:
[accuracy report]
EFFECTIVENESS SUMMARY:
| Artifact | Effectiveness | Recommendation |
[table]
PROPOSED IMPROVEMENTS:
[list with PDCA format]
CYCLE LOGGED: [count] artifacts updated
Run after the weekly harvest — the harvest provides fresh data for the compounding cycle.
Add to Streeling workflow or run manually:
CronCreate("47 13 * * 5", "/demerzel compound")
(Runs Fridays at 8:47 AM EST, 30 minutes after the harvest)
/demerzel promote for promotion proposals, /demerzel evolve for evolution insightsDemerzel CANNOT approve her own constitutional changes. Per the mandate:
The compounding cycle is the primary writer of evolution state. It MUST:
state/evolution/ — this is the inputstate/beliefs/ for contextstate/pdca/ for improvement trackingstate/evolution/ — add new events, update metrics, refresh assessmentsstate/pdca/ for each proposed improvementstate/beliefs/ if the cycle revealed new truthsEach cycle's output is the next cycle's input. The evolution log grows richer over time:
logic/governance-evolution.schema.json, policies/scientific-objectivity-policy.yaml (confidence calibration), policies/kaizen-policy.yaml (waste taxonomy, PDCA), constitutions/demerzel-mandate.md Section 4 (accountability)