5 projects
crucible-harden
Adversarial test-hardening loop for AI-built code: Tester vs Critic closing mutation feedback, mechanical verdicts, metered spend, receipts.
oracle-gate
Reference CLI for The Oracle Gate — a framework for testing AI-built code. Model-agnostic cross-model review + per-tier evidence checklist.
agent-cost-attribution
Per-stage token/cost attribution and silent-degradation detection for multi-agent workflow runs (stdlib-only).
guarded-rag
Guarded RAG: grounded answers, refuse-when-unsupported, PII redaction, and an eval harness with metrics. Stdlib core, bring-your-own model.
mcp-agent-gate
An MCP server that lets an AI agent gate its own work: deterministic checks, refute-first review, and tamper-evident honest receipts. Fleet Mode, as a tool.
