How do I install the AUDIT-skill-atlas skill?

Run npx skillmds@latest add ariffazil/audit-skill-atlas-2 in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

What does the AUDIT-skill-atlas skill do?

Unified skill inventory, gap detection, and cross-cutting orchestration for AAA catalog + Grok/Claude/Codex/Hermes/Kimi/OpenClaw harness views. The mesa above the terrain — see the whole, find the missing, route the right skill. Load when starting a new task, auditing skill health, unifying CLI agent skills, onboarding a new organ, or when you don't know which skill to load. It is listed under AI & ML on SkillMD.

Is AUDIT-skill-atlas safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with AUDIT-skill-atlas?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is AUDIT-skill-atlas free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published AUDIT-skill-atlas?

ariffazil (@ariffazil) published this skill. Their other Agent Skills are listed on their SkillMD profile.

AUDIT-skill-atlas by ariffazil — Agent Skill

Name: AUDIT-skill-atlas
Rating: 4.5 (4 reviews)
Author: ariffazil

META-MESA — Skill Atlas & Gap Detection

DITEMPA BUKAN DIBERI — Intelligence is forged, not given. The mesa above the terrain. See the whole. Find the missing. Route the right.

What This Skill Is

This is the meta-skill — a skill about skills. It provides:

Inventory — live multi-surface counts (not a frozen “35”)
Gap detection — missing, thin, dual-named, harness-divergent
Routing — given a task and harness, which skill(s) to load
Health scoring — freshness, collision, catalog vs view drift
Cross-harness unification — AAA catalog ↔ CLI agent views

It does NOT execute, judge, or seal. It classifies, routes, and illuminates.

Iron rule: AAA is the catalog. Harnesses are views. Do not invent parallel catalogs.

Ops skill: skill-unification (AAA + mesh views) — alias/mesh/boot contract.

Live atlas (T₁): /root/A-FORGE/forge_work/2026-07-12/GROK-CLI-AAA-SKILL-UNIFICATION-ATLAS.md
JSON: …/GROK-CLI-AAA-SKILL-UNIFICATION-ATLAS.json

§0. MULTI-HARNESS UNIFICATION (Grok × CLI × AAA) — 2026-07-12

Architecture

BOOTSTRAP_MANIFEST (9 universals)     ← always first
        ↓
AAA/skills  (catalog + V3 registry)   ← sole named truth
   +  .agents/skills (doctrine/stage) ← shared federation core
        ↓ symlink mesh
~/.grok | ~/.claude | ~/.codex        ← views (+ harness-native)
        ↓ separate trees
Hermes categories | Kimi roles | OpenClaw owned

Live inventory (OBSERVED T₁)

Surface	Count	Role
Grok `~/.grok/skills`	116 resolvable	View: AAA 67 + agents 37 + native 12
AAA `/root/AAA/skills`	70 top-level + substrate/knowledge	Catalog
V3 registry	63 logical (6+3+54)	Short names
`.agents/skills`	43–50	Stage/domain doctrine
Claude / Codex	104 / 107	Same mesh (symlink)
Hermes	180 SKILL.md / 34 cats	Category tree — not flat mesh
Kimi	7 role contrast skills	Local, not catalog
Legacy meta-mesa claim	37	STALE — do not use

Grok native keepers (harness-only)

arif-governed-autonomous-execution · grok-zen-aaa-substrate · grok-federation-skill-upgrader · orthogonal-skill-update · create-skill · check-work · help · imagine
Pruned (2026-07-12): docx · pptx · xlsx · code-review → ~/.grok/skills/.deprecated/

Dual-name gap — SEALED (2026-07-12)

Machine table (63/63 RESOLVED): /root/AAA/skills/SKILL_ALIAS_TABLE.json
Human: A-FORGE/forge_work/2026-07-12/SKILL_ALIAS_TABLE.md
Hermes bridge: A-FORGE/forge_work/2026-07-12/HERMES-V3-DOMAIN-BRIDGE.md
Mesh sync: bash /root/AAA/skills/scripts/skill-mesh-sync.sh [--apply|--check]

V3 short	Primary disk path
`kernel-bind` … `audit-seal`	`AAA/skills/substrate/<name>`
`know-*`	`AAA/skills/knowledge/<name>`
`meta-atlas`	`meta-mesa-skill-atlas`
`geo-*`	`geox-*` / `geological-artifact-rigor` / `prospect-maturation-workflow`
`wealth-thermo`	`wealth-capital-thermodynamics`
`forge-verbs` / `forge-exec`	`kernel-verbs-aforge-hands`
(full list)	see JSON

Unification phases — status 2026-07-12

AUTHORITY — ✅ AAA + V3 sole catalog
ALIAS TABLE — ✅ 63/63 sealed JSON
PROFILES — ✅ grok + opencode in V3
META-MESA REFRESH — ✅ v1.2.0+
MESH SYNC — ✅ script + applied (substrate/knowledge links)
HERMES BRIDGE — ✅ category→domain doc
PRUNE — ✅ Grok docx/pptx/xlsx/code-review → .deprecated
BOOT GATE — ✅ documented below + BOOTSTRAP_MANIFEST

BOOT GATE (all CLI agents)

BIND → GROUND → ROUTE → RECALL → VERIFY → SEAL → KNOW → READY

Only after READY may domain skills load. Source: AAA/skills/BOOTSTRAP_MANIFEST.json (9 universals).

Harness load packs

CLI	Always	On demand
Grok	BOOTSTRAP 9 + `arif-governed` + `grok-zen-aaa-substrate`	Domain via this atlas / V3
Claude / Codex	BOOTSTRAP 9 + mesh	Same domain pack
Hermes	`arifos-auto-init` + seven-zen + federated-skill-architecture	Category skill
OpenClaw	openclaw-agentic + memory	ops / a2a
Kimi	role contrast skills	forge / dev

Anti-patterns (multi-harness)

Anti-pattern	Remedy
Copy skill bodies into every `~/.X/skills`	Symlink to AAA / .agents
Second “Grok catalog” of 100+ natives	Keep ≤12 harness keepers
Flatten Hermes into AAA	Bridge map only
Route by V3 short name without path	Resolve via alias table
Trust meta-mesa “37 total”	Re-probe disk / use §0

§1. THE SEVEN STAGES (Constitutional Pipeline)

The golden path. Every governed action flows through these stages.

#	Skill	Stage	Role	Risk	Key Output
000	`000-init-intent-classify`	1	Entry gate	HIGH	loop_class, required_organs, blast_radius, next_lawful_call
111	`111-sense-evidence-observe`	2	Evidence binding	HIGH	evidence_table, epistemic_tags, contradiction_scan, gaps
333	`333-mind-plan-generate`	3	Plan generation	HIGH	3+ candidate_plans, falsification_checks, DAG
666	`666-heart-critique-stress`	4	Consequence scan	CRITICAL	risk_register, dignity_check, posture_recommendation
888	`888-judge-verdict-render`	5	Verdict	CRITICAL	SEAL / SABAR / HOLD / VOID
010	`010-forge-execute-warrant`	6	Execution	CRITICAL	execution_receipt, rollback_path
999	`999-vault-seal-immutable`	7	Memory	HIGH	VAULT999 record, lineage, lessons

Stage Gaps (What Falls Between)

Gap	Between	Impact	Mitigation
No priority queue	000 → 111	Multiple intents compete, no ordering	Manual triage; future: session-state priority field
No evidence freshness decay	111	Evidence ages but no re-observation trigger	Use `arif_observe` freshness_weight on recall
No cost estimation	333	Plans don't estimate compute/API/human cost	Add cost column to plan steps manually
No plan revision loop	666 → 333	SABAR/HOLD doesn't auto-revise plan	888 judge can route back to 333
No temporal cascade modeling	666	Consequences are static snapshots	Use `wealth_monte_carlo_simulate` for temporal
No verdict expiration	888 → 010	Old SEAL could authorize stale execution	Check seal timestamp at 010
No partial execution handling	010	Step 3/5 failure = undefined state	Define per-plan partial-rollback strategy
No vault retrieval	999	Write-only from skill perspective	Use `arif_vault_query` for read path

§2. DOMAIN SKILLS BY ORGAN (legacy inventory — prefer §0 + SKILL_ALIAS_TABLE.json)

GEOX — Earth Intelligence (7 skills, THICK coverage)

Skill	Purpose
`geox-constitution`	Constitutional floors, epistemic style, HOLD triggers
`geox-claim-grammar`	Claim YAML schema, multi-hypothesis mandate
`geox-earth-evidence`	Evidence discipline, artifact refs, handoff protocol
`geox-epistemic-ladder`	7-rung ladder, iron rule (lower beats higher)
`geox-petrophysics-bounds`	Vsh, porosity, Sw, AI, permeability QC
`geox-contradiction-engine`	7 contradiction types, multi-hypothesis scanner
`geox-redteam-hantu`	Anti-hallucination, F9 enforcement

WEALTH — Capital Intelligence (4 skills, GOOD coverage)

Skill	Purpose
`wealth-capital-thermodynamics`	13 primitives grounding all WEALTH computation
`wealth-capital-reasoning`	POS, EMV, valuation, governed consequence
`wealth-collapse-signature`	7 institutional failure patterns
`wealth-law-anthropology`	Malaysian law: faraid, KTN, MA63, Syariah, adat

Infrastructure (8 skills)

Skill	Domain	Purpose
`mcp-mastery`	MCP	Trust surface, protocol, tool forging, transport
`mcp-apps-builder`	MCP	SEP-1865 interactive HTML UI in conversations
`webmcp-site-builder`	MCP	WebMCP sites, W3C origin trial
`github-operations`	Git	Issue triage, PR review, CI diagnostics
`federation-observability`	Ops	OTEL, Prometheus, Grafana, LGTM
`hf-mastery`	ML	Governed HuggingFace: hub, datasets, inference, fine-tune
`iron-shell-render`	Web	Three-layer site architecture
`forge-document-intelligence`	Document	EMD stack (Encode-Metabolize-Decode) — VLM perception, provenance, governed action

Forge & Agent (3 skills)

Skill	Purpose
`aforge-execution`	Build, deploy, verify. Gated by SEAL.
`agentic-builder`	Build governed agents: identity, skills, tools
`skill-creator`	Bootstrap, design, package new skills

General / Meta (4 skills)

Skill	Purpose
`apex-theory`	Physics + math + symbolic code + meaning
`entropy-thermo-zen`	Thermodynamic entropy management (TZQ)
`agentic-fitness-law`	Fitness law, extinction pressures, immune system for tools/agents
`symbolic-order-trust-architecture`	Structural trust mapping

Symbolic Order (2 of 5 planned)

Skill	Purpose
`symbolic-order-collective-bias`	Detect population-level bias via naming game
`symbolic-order-trust-architecture`	Trust symmetry violations

WELL — Human Readiness (1 skill, THIN)

Skill	Purpose
`well-substrate-readiness`	Vitality, fatigue, dignity. Reflect only.

AAA — Control Plane (1 skill, THIN)

Skill	Purpose
`aaa-cockpit`	Agent registry, session cockpit, A2A gateway

A2A / Federation (1 skill)

Skill	Purpose
`a2a-federation-builder`	Agent Cards, JSON-RPC /a2a, federation directory

§3. GAP REGISTER — Missing Skills

3A. Referenced But Do Not Exist (Phantom Skills)

Cleaned 2026-07-03: Phantoms replaced with actual tool alternatives. Remaining = PLANNED skills only.

Missing Skill	Status	Function
`070-lock-humility-godel`	✅ CLEANED → `arif_think` mode=verify + `666-heart-critique-stress`	Final epistemic gate before SEAL-grade claims
`034-dag-plan-orchestrate`	✅ CLEANED → `333-mind-plan-generate`	Multi-step execution graph for cross-organ work
`444-route-organ-direct`	✅ CLEANED → `arif_route` + `tools-embodiment-application`	Organ routing and direct addressing
`950-seal-vault-audit`	✅ CLEANED → `arif_seal` mode=verify	Vault999 seal verification and chain audit
`geox-basin-interpreter`	✅ CLEANED → `geox_basin` directly	Upstream O&G basin interpretation for Petronas
`090-check-tool-preflight`	✅ CLEANED → `666-heart-critique-stress`	Tool preflight check
`symbolic-order-solidarity-monitor`	🔵 PLANNED	Calhoun Phase C population-level detection
`symbolic-order-mythic-attestor`	🔵 PLANNED	Sacred domain violation detection
`symbolic-order-layer-governance`	🔵 PLANNED	L1/L2/L3 hierarchy enforcement

3B. Entirely Missing Domains (No Skill Exists)

Missing Domain	Organ	Impact	Priority
arifOS kernel operations	arifOS	No domain skill for session lifecycle, world model, kernel internals	P2
VAULT999 deep operations	arifOS	No receipt/chain/hash management skill	P3
HERMES/ASI Telegram agent	555-ASI	No skill despite SOUL.md existing at /root/.hermes/SOUL.md	P3
Docker/container management	Infrastructure	VPS has 17 hostinger tools, no container skill	P4
DNS/Cloudflare management	Infrastructure	No skill for DNS, Workers, R2, Pages	P4
Database operations	All organs	No PostgreSQL/Supabase schema management skill	P4
WEALTH stock analysis deep	WEALTH	12 modes of wealth_stock_analysis, no dedicated skill	P3
WEALTH personal finance	WEALTH	D1 personal scope, no dedicated skill	P4
~~Document intelligence~~	~~A-FORGE~~	~~Filled 2026-07-02 — `forge-document-intelligence` forged~~	✅ FILLED

3C. Thinly Covered Organs

Organ	Current Depth	What's Missing
WELL	1 skill	Sleep science, fatigue modeling, cognitive depletion, metabolic flux, dignity enforcement patterns
AAA	1 skill	A2A mesh routing details, cockpit React dashboard, deliberation integration, agent lifecycle
Symbolic Order	2 of 5	solidarity-monitor, mythic-attestor, layer-governance — cluster is incomplete

§4. ROUTING TABLE — Which Skill For Which Task

Given a natural-language intent, this table routes to the correct skill.

Intent Pattern	Skill to Load	Why
"plan this" / "govern this" / "classify intent"	`000-init-intent-classify`	Stage 1 entry gate
"observe" / "gather evidence" / "what do we know"	`111-sense-evidence-observe`	Stage 2 evidence binding
"make a plan" / "generate options" / "how should we"	`333-mind-plan-generate`	Stage 3 planning
"stress test" / "what could go wrong" / "critique"	`666-heart-critique-stress`	Stage 4 consequence scan
"judge" / "render verdict" / "is this lawful"	`888-judge-verdict-render`	Stage 5 verdict
"execute" / "forge" / "build this"	`010-forge-execute-warrant`	Stage 6 execution
"seal" / "record" / "vault this"	`999-vault-seal-immutable`	Stage 7 memory
"seismic" / "well log" / "petrophysics" / "basin"	`geox-constitution` + domain geox-*	Earth intelligence gate
"NPV" / "IRR" / "capital" / "investment"	`wealth-capital-thermodynamics`	Capital primitives first
"collapse" / "institutional failure" / "Enron"	`wealth-collapse-signature`	Failure forensics
"Malaysian law" / "faraid" / "pusaka" / "MA63"	`wealth-law-anthropology`	Legal domain
"sleep" / "fatigue" / "vitality" / "dignity"	`well-substrate-readiness`	Human readiness
"build MCP tool" / "forge tool" / "MCP server"	`mcp-mastery`	MCP infrastructure
"build a website" / "iron shell" / "static site"	`iron-shell-render`	Site architecture
"GitHub PR" / "CI broken" / "issue triage"	`github-operations`	Git operations
"build an agent" / "new agent identity"	`agentic-builder`	Agent construction
"create a skill" / "new skill"	`skill-creator`	Skill creation
"entropy" / "cleanup" / "system chaos"	`entropy-thermo-zen`	TZQ framework
"fitness" / "tool evolution" / "why tools survive"	`agentic-fitness-law`	Fitness law + extinction pressures
"APEX" / "consciousness" / "physics of AI"	`apex-theory`	Theoretical framework
"A2A" / "agent communication" / "federation mesh"	`a2a-federation-builder`	Inter-agent surface
"bias" / "collective behavior" / "population"	`symbolic-order-collective-bias`	Social-symbolic
"trust" / "trust architecture" / "trust violation"	`symbolic-order-trust-architecture`	Trust mapping
"observability" / "tracing" / "metrics"	`federation-observability`	LGTM stack
"HuggingFace" / "model hub" / "fine-tune"	`hf-mastery`	ML governance
"PDF" / "OCR" / "scan" / "document" / "extract text"	`forge-document-intelligence`	EMD stack — VLM perception + provenance + governance
"what skill should I load" / "skill gap" / "skill audit"	THIS SKILL (meta-mesa)	Meta-routing

§5. HEALTH SCORING — Skill Freshness

Run periodically to detect stale skills.

Per-Skill Health Check

skill_health = {
    "name": str,           # skill name
    "version": str,        # from frontmatter
    "forged_date": str,    # from frontmatter
    "days_since_forge": int,
    "references_count": int,   # how many other skills reference it
    "phantom_refs": int,       # references to skills that don't exist
    "has_prompt": bool,       # prompts/ directory exists
    "has_test": bool,         # test coverage exists
    "organ_coverage": float,  # 0.0-1.0 depth for its organ
}

Federation Skill Health Summary

federation_skill_health = {
    # T₁ 2026-07-12 — re-probe; do not trust frozen integers without fs check
    "v3_logical_skills": 63,      # AAA FEDERATED_SKILLS_REGISTRY_V3
    "aaa_top_level_dirs": 70,     # with SKILL.md (excl. deep archive noise)
    "grok_resolvable": 116,       # symlink mesh + 12 native
    "claude_resolvable": 104,
    "codex_resolvable": 107,
    "agents_core": 43,
    "hermes_skill_md": 180,       # category tree, not mesh
    "dual_name_debt": False,      # alias table 63/63 sealed + audited
    "body_homes": ["AAA/skills", ".agents/skills"],  # doctrine_core dual body — honest
    "thin_organs": ["WELL", "SYMBOLIC"],
    "thick_organs": ["GEOX", "OPS", "META"],
    "unification_maturity": 0.92, # audited multi-agent 2026-07-12
    "collision_risk": float,
    "stale_skills": int,
    "coverage_score": float,
}

Stale Threshold

Days Since Forge	Status	Action
0-7	FRESH	No action
8-14	CURRENT	Review on next session
15-30	AGING	Verify references still valid
30+	STALE	Audit needed, consider archive or refresh

§6. CROSS-CUTTING CONCERNS — The Gaps Between Skills

These are capabilities that NO single skill covers but the federation needs.

6.1 Evidence Lifecycle (covers 111, but not full lifecycle)

Phase	Covered By	Gap
Collection	111-sense-evidence	✅
Tagging	111-sense-evidence	✅
Contradiction detection	111 + geox-contradiction-engine	✅
Freshness decay tracking	NONE	❌ No skill tracks evidence aging
Cross-session correlation	NONE	❌ No skill correlates evidence across sessions
Evidence provenance chain	111 (partial)	⚠️ Only table row, not full chain of custody
Evidence retirement	NONE	❌ No skill decommissions stale evidence

6.2 Plan Lifecycle (covers 333, but not full lifecycle)

Phase	Covered By	Gap
Generation	333-mind-plan	✅
Multi-hypothesis	333-mind-plan	✅
Revision after critique	NONE	❌ No 333↔666 feedback loop
Cross-plan conflict detection	NONE	❌ No concurrent plan conflict detection
Cost estimation	NONE	❌ No compute/API/human cost per step
Plan versioning	NONE	❌ No plan history tracking
Plan → execution handoff	010-forge (consumes)	⚠️ No explicit plan_id passing

6.3 Operator Readiness (covers WELL, but only at stage 666)

Check Point	Currently	Gap
Before planning (stage 333)	❌ No WELL check	WELL should gate plan generation
Before execution (stage 010)	❌ No WELL check	WELL should gate all T2/T3 actions
During critique (stage 666)	✅ 666 integrates WELL	Only place WELL is checked
Post-execution (after 010)	❌ No WELL check	Should verify operator wasn't overloaded

6.4 Social-Symbolic Dimension (AGENTS.md declares it, no skill operationalizes it)

Aspect	Status
Collective bias detection	✅ `symbolic-order-collective-bias`
Trust architecture mapping	✅ `symbolic-order-trust-architecture`
Solidarity monitoring	❌ Missing skill
Mythic domain violation	❌ Missing skill
Layer governance (L1/L2/L3)	❌ Missing skill
Institutional dynamics modeling	❌ Not operationalized anywhere
Population behavior dynamics	⚠️ Partial in `wealth_monte_carlo_simulate` population_mode

§7. ANTI-PATTERNS

Anti-Pattern	Description	Remedy
Skill overload	Loading 5+ skills for a simple task	Use this meta-skill to identify the MINIMUM skill set
Phantom reliance	Referencing a skill that doesn't exist	Run gap register (§3A) and forge missing skills
Organ mismatch	Loading a GEOX skill for a WEALTH task	Check routing table (§4)
Stage skipping	Going straight to 010 without 000→111→333→666→888	Always start with 000-init-intent-classify
Single-skill tunnel	Using only one skill when the task spans organs	Load the cross-cutting concern from §6
Stale skill	Using a skill forged 30+ days ago without verification	Check health score (§5)
Collision blindness	Two skills covering the same capability differently	Run collision detection on overlap

§8. QUICK REFERENCE — Minimal Skill Set Per Task Type

Task Type	Minimum Skills	Maximum Skills (if deep)
Simple observation	000	000 + 111
Code edit/build	000 + 010	000 + 111 + 333 + 010 + 999
Capital analysis	000 + wealth-capital-thermodynamics	000 + 111 + wealth-capital-thermodynamics + wealth-capital-reasoning + 666 + 888 + 999
Geological interpretation	000 + geox-constitution	000 + 111 + geox-constitution + geox-earth-evidence + geox-epistemic-ladder + geox-contradiction-engine + 666 + 888 + 999
Federation audit	000 + entropy-thermo-zen	000 + entropy-thermo-zen + this meta-skill + 999
Tool / alias / fitness decision	agentic-fitness-law	000 + agentic-fitness-law + entropy-thermo-zen + 999
Build new agent	000 + agentic-builder	000 + agentic-builder + skill-creator + a2a-federation-builder
Skill gap detection	This meta-skill	This meta-skill + skill-creator
Constitutional judgment	000 + 888	000 + 111 + 333 + 666 + 888 + 999
Infrastructure change	000 + aforge-execution	000 + 111 + 333 + aforge-execution + mcp-mastery + 010 + 999
Document processing / OCR / PDF extraction	000 + forge-document-intelligence	000 + 111 + forge-document-intelligence + 666 + 888 + 999 (if high-stakes)

§9. RECOMMENDATIONS — Priority Queue

2026-07-12: Prefer aliases, profiles, prune, mesh-sync over forging more skills.
Many §3B “missing domains” are now filled under long names (docker-, vault999-, infra-guardian, hermes categories).

Priority	Action	Status / Why
P0	Alias table short→path	✅ 63/63 + multi-agent audit
P0	V3 `grok` + `opencode` profiles	✅
P0	Mesh sync + check clean	✅
P0	Grok office prune	✅ `.deprecated`
P0	Hermes bridge (no flatten)	✅
P0	Tombstone dual-state normalize	✅ RESTORED_LIVE after auditor CAUTION
P1	Migrate high-traffic `.agents` geo/wealth bodies into AAA (optional)	Open — honesty already documented as doctrine_core
P2	`symbolic-order-solidarity-monitor`	Open
P3	`well-fatigue-science`	Open
P3	Rewrite residual §2 phantom long-names to V3	Open (use alias table / §0 until then)

§10. TELEMTRY

{
  "skill_name": "meta-mesa-skill-atlas",
  "version": "1.2.0",
  "trigger_phrase": "{{trigger_phrase}}",
  "selected_reason": "{{selected_reason}}",
  "v3_logical": 63,
  "grok_resolvable": 116,
  "aaa_catalog_top": 70,
  "dual_name_debt": false,
  "unification_maturity": 0.92,
  "atlas_artifact": "A-FORGE/forge_work/2026-07-12/GROK-CLI-AAA-SKILL-UNIFICATION-ATLAS.md",
  "routing_decision": "{{routed_to_skill}}",
  "latency_ms": 0
}

§11. RECURSIVE SCORECARD

Metric	Target	Notes
Activation Precision	> 0.95	Don't load for domain tasks
Routing Accuracy	> 0.90	Correct skill for intent and harness
Gap Detection Rate	> 0.85	Find missing + dual-named
Phantom Resolution Rate	1.00	Resolve via alias table, never invent path
Doc Freshness	1.00	Re-probe disk; atlas in forge_work/
Catalog/View drift	0	Grok/Claude/Codex mesh == AAA+.agents

Forged: 2026-06-28 by FORGE (000Ω) for F13 SOVEREIGN
Updated: 2026-07-02 — forge-document-intelligence filled
Updated: 2026-07-12 — multi-harness §0; live inventory; V3/bootstrap alignment; atlas artifact
Canon: AAA catalog · harness views · BOOTSTRAP first
DITEMPA BUKAN DIBERI — The mesa sees the terrain. The terrain does not see the mesa.

AUDIT-skill-atlas

File contents