aq217_audit_agentlint_8laws_and_forgetting_0call)Audits AGENTS.md, SKILL.md, DESIGN.md, and BENCHMARK.md bundles in Store S63 against 8 AgentLint laws including paired bare-baseline lift and epistemic forgetting hygiene.
Token-budget, smell & paired-lift static auditor (0 model calls)
Substrate: file_cas · p99 Claim Target: ≤ 6 ms · Phase: 201 - 220 V42..V50 Epistemic Forager, FF1 FPE, BotFabric Homeostat, R-WoM/Futarchy, GoodNodes & SoftRules
Context-bloat surgical trimmer & skill benchmark author
5-Role Component Pod: 0-Call Slot / Solver + OpenSearch / Split-Supervisor (≤4k tok) / Diagnostic Repair + Twin Shim / CPOP + Sister Distiller.
rux_harness_benchmark_auditCommit Invariant: Enforces identical tool/prompt scaffold parity across models on uncontaminated holdouts and flags scaffold inflation whenever retry budgets or custom wrappers diverge.
Architectural Problem Solved: Separates genuine model or substrate lift from benchmark-specific scaffold gaming.
POST https://api.teleon.dev/v1/tenants/:tenantId/queues/aq217_audit_agentlint_8laws_and_forgetting_0call/enqueue
{
"queue_category": "rux_harness_benchmark_audit",
"priority": 80,
"payload": { "stage": "217", "objective": "AgentLint 8-Law & Forgetting Audit" }
}