agent-audit

sofagent is an open-source FDE Harness for AI coding agents — on entry, write your business judgment into files (workflows, ontology, AI nodes); after departure, audit every change against them. 25 codified audit rules, tamper-evident chain, snapshots, and an MCP toolset (104 tools, 11 plugins). Everything can be FDEing.

Reproducible research on verification boundaries, plus the earlier Nobulex receipt prototype. Current financial decision-integrity gateway: github.com/arian-gogani/nobulex-registry.

Oh! My Agent Teams - VALP protocol for visible, evidence-backed multi-agent work.

Java-native runtime for long-running AI agents with live execution control, human approvals, and audit trails.

Cryptographically verifiable audit trail for AI systems — Ed25519 signed, hash-chained Decision Receipts with ISO 42001 / NIST AI RMF compliance. No blockchain, no SaaS.

The open specification and reference SDK for recording AI Agent decision chains. Every decision, recorded. Every alternative, documented.

CML (Causal Memory Layer) — a foundational memory layer for recording reasons, permissions, and responsibility behind actions, not just events or results. Enables systems in AI, fintech, security, and distributed computing to preserve meaning and causal accountability across time, independent of execution or transport.

Pi extension that audits rpiv-todo board drift every 10 agent loops using TypeSafe's jev model, injecting corrective nudges when the agent wanders off-task

Claude Code Skill that audits AI agents using systems thinking — leverage points, feedback loops, emergent behavior, and paradigm/mental model — to find structural risks code review misses.

TreLLM is an LLM-driven trading audit and live-readiness control system; TradeArena is its public leaderboard for replayable trajectories, risk gates, execution calibration, and reproducible agent evidence.

Forward-deployed AI implementation toolkit: customer context, readiness scoring, audit trails, rollout checks, and MCP patterns.

Claude Code skill for static audit of OpenClaw config and agent proposals — llms.txt-first research, F1–F12 incident library, Notion-bridged Galatea audit workflow.

상담 로그를 ALF 호환 테스트 케이스로 바꾸고, 지식 변경 전후의 해결률과 정책 위반을 회귀 비교해 배포 여부를 권고하는 감사 파이프라인

Proof-first tools for AI agents. So humans stay accountable when machines fail.

Built on affaan-m/ECC by @affaan-m: a Claude Code skill that finds agent and LLM app memory, tool, wrapper, and output bugs.

Does an agent's action follow from the evidence it actually got? A ternary evidence-to-action check for MCP traces.

Open-source CLI that audits AI agents for security vulnerabilities, performance issues, and model fit — based on OWASP Agentic AI Top 10

Audits configured AI agents for task success, tool judgment, authority, truthful completion, memory isolation, recovery, and release readiness using observable evidence.