agent-traces

The living ecosystem where AI agents complete tasks through workflow loops, improve through iterative execution, are evaluated by mentor agents or humans in the loop, and turn completed work into reusable work experience and data to improve future agents.

Observability for AI agents. See what your agent did, why it cost that much, and what to fix.

Use cultivar to test your Agent Skills and Docs by running them in sandboxes, and across different agents.

Compile an AI agent's repeated workflows into deterministic, auditable routines that replay for free, with a fallback to the agent.

Generate realistic multi-agent workflow traces with LLM-enriched content, semantic validation, and PM4Py compatibility. pip install open-agent-traces

Trace Intelligence for Claude Code, Codex, Cursor, OpenCode, Grok and Droid sessions.

Agent serving trace from https://freeinference.org

A governed learning layer for AI agents — turns execution traces into reviewed memories, reusable skills, and evidence-backed training data.

Agent Trace Analysis Platform (atap) — pluggable pipeline for trace representation, analysis, error classification, failure attribution and recovery for LLM agents

TRACES.md — AGENTS.md, but for what agents did. A simple, open format for recording what AI agents did, for error analysis and self-optimization.

📚 Curated catalog of agent-training datasets + a toolkit that normalizes, deduplicates, and quality-tiers them into one schema. Produces 🤗 voidful/agent-sft.

AgentDossier - A vendor-neutral registry of AI agents that shows what compliance and trust evidence exists for each Agent (HIPAA, SOC 2, FedRAMP, ISO42001, EU AI Act), Traces and CVEs. Agents are discovered through the open standards they publish (ARD, A2A, MCP) as well as public registries, and same engine can scan enterprise's own network.

TraceCC gives AI agents a CLI-native long-term memory over their own session history. Searchable, scoped, and sized for context windows.

Open-source visual debugger for coding agents: see what the agent reported doing and what actually changed.

Accepted to NeurIPS 2026 (Main Track). Reviewer precision and critique uptake are separable: a multi-agent protocol can detect errors well and still fail to repair them. Code, data and analysis for arXiv:2607.15388.

Unroll: read agent traces and chat logs in VS Code. Claude Code, Codex and other JSONL conversations as a readable chat. Public docs, examples and releases; implementation source is private.

Collect, normalize, audit, and render coding-agent traces in one portable format.

Paired red and blue cyber-range agent traces

Retry Safety Analyzer: deterministic HIGH/MEDIUM/SAFE when a side-effecting tool may have been retried. Offline import — never executes the trace. HIGH is risk, not proof. External validation open.

Reference implementation, benchmarks, and traces for "Parsing the Stream: A Live Trace Model for Long-Horizon Agents and Their Observers" (arXiv:2609.01466). Append-only trace ledger → deterministic run state → per-consumer views for the human observer and the agent.