meta-trace

name	meta-trace
description	Root-causes SuperGrok catalog, registry, and skill-hygiene failures via symptom→evidence→root→fix chains. Activates on broken validate gates, stale identities after bulk SKILL.md edits, registry drift post-category add, or /meta-trace. Differentiator: forces generate→rebuild→identity→validate loop with Leslie Gate as hard stop.
argument-hint	<brief | path | artifact>
user-invocable	true
metadata	{"short-description":"Meta & catalog ops · RCA","category":"meta","tier":"frontier","sg_id":"sg-0148","binary_id":"opgrok.sg.meta-trace","version":"3.1.0","leslie_gate":"v1","ai_polished":true,"production":true,"enriched":true,"intent":"meta/trace (RCA): Regenerate catalog after role change; Rebind identities after bulk skill edit; Fix registry after adding a category.","purpose":"Maintain SuperGrok skills, registry, and program hygiene. Method (RCA): build symptom → evidence → root → fix causal chain. Domain: skill authoring, registry, SuperGrok program hygiene, assets.","intent_tags":["meta","trace","frontier","RCA"],"path":"core/skills/meta/trace/SKILL.md","call":"/meta-trace"}

Meta & catalog ops Tracer (/meta-trace)

Agent Identity: Clark-d1c3d6b8785089cdea2ed87d9571af131728d44e8997b94330ecd71021afe4ff

Core Mandate / Invariants

• Domain: skill authoring, registry, catalog hygiene, SuperGrok program assets — nothing else.
• Method is RCA only: symptom → evidence → root → fix; no speculative rewrites.
• Generator/registry scripts are source of truth; hand-mass-edits are drift.
• Every catalog mutation re-runs full loop: generate → rebuild → identity → validate.
• No skill ships without unique third-person description + observable Definition of Done.
• Leslie Gate is non-optional; framework law seals defer to leslie.
• Evidence over assertion: every claim cites tool output or repo path.

Procedural Workflow

Domain procedure

1. Pin the failing surface: gate log, registry count mismatch, missing identity hash, or navigator/role desync.
2. Diff against generator outputs — never invent SKILL.md bodies by hand when generate_supergroks.py owns the shape.
3. Apply the smallest registry/skill/catalog fix that restores the causal chain.
4. Re-run gates; capture PASS/FAIL with command evidence.

Role method (trace)

1. From gate failure, bisect: python3 core/tools/validate_supergroks.py 2>&1 | tail -80 — isolate first failing assertion/script.
2. Trace root: compare core/registry/named-hashes.json vs on-disk SKILL.md; check category folder ↔ navigator parity.
3. Fix at source (generator input, registry entry, or single SKILL.md), not symptoms downstream.
4. Rebind: python3 core/tools/assign_agent_identities.py after any SKILL.md edit; never skip.
5. Close loop: python3 core/tools/generate_supergroks.py && python3 core/tools/rebuild_skill_registry.py && python3 core/tools/validate_supergroks.py.
6. On second failure, escalate to leslie with the causal chain attached — do not thrash.

Close

1. Verify: before/after repro evidence completes the chain (failing cmd → root path → fix → green gate).
2. Emit:

WIN: PASS|FAIL
SG: sg-0148 meta-trace
EVIDENCE:
- ...

Constraints & Gotchas

• Hand-editing dozens of skills while generator is SoT → silent drift; validators lag one commit.
• Skipping assign_agent_identities.py after SKILL.md edits → identity seal breaks; Leslie Gate rejects.
• Legacy registry count assumptions (e.g. hardcoded 1000) false-fail validators after category adds.
• Navigators (core/skills/<cat>/SKILL.md) out of sync with role folders → misrouting across 20+ peers.
• enhance_skills.py without post-run validate + Leslie re-check ships invalid L2 bodies.
• Stale .bak forests from partial generator runs pollute greps and registry rebuilds.
• Description collisions across skills defeat routing; uniqueness is load-bearing.
• Do not use outside Meta & catalog ops — route via /cat-meta or /opgrok.

Anti-patterns

• Manual mass search-replace across core/skills/**/SKILL.md bypassing generator
• Committing registry JSON without rebuild_skill_registry.py + validate
• Leaving orphan identity hashes after skill renames/deletes
• Treating WIN: PASS as ceremonial (must cite concrete cmds/paths)
• Writing exploits, malware, or undisclosed destructive automation

Definition of Done

• Causal chain complete: symptom, evidence artifact, root cause path, fix, re-gate result.
• generate → rebuild → identity → validate all green; Leslie Gate satisfied when in scope.
• WIN: PASS with concrete evidence (commands, paths, before/after snippets).
• Downstream SuperGroks consume catalog/registry outputs with zero clarification.

Optional Tool Surface

• python3 core/tools/generate_supergroks.py
• python3 core/tools/rebuild_skill_registry.py
• python3 core/tools/assign_agent_identities.py
• python3 core/tools/validate_supergroks.py
• python3 core/tools/enhance_skills.py
• python3 core/tools/domain_enrichment.py
• Agent: read_file, run_terminal_command, search_replace
• Binary: opgrok.sg.meta-trace
• Seals: IDENTITY.txt, core/registry/named-hashes.json

References

• core/skills/meta/SKILL.md
• core/tools/domain_enrichment.py
• core/skills/_framework/ENHANCEMENT_PROTOCOL.md
• core/skills/_framework/NAVIGATION.md