Agent Replay Model Card

这是可索引项目报告证据页：它保留 Agent Replay Model Card 的源材料入口、机制线索和限制提醒；正文仍需 reader/editor 与 academic public-copy review 后才能当作最终结论引用。

Agent Replay Model Card

1. Role in Self Evolve

Agent Replay is a local-first desktop evals, observability and AI memory stack for coding agents, with traces, embedded storage, vector memory, MCP/Claude Code integration and no cloud dependency.

2. Working Principle

local traces -> evals -> memory retrieval -> agent debugging

3. Evidence Path

web search/public GitHub page observed agentreplay/agentreplay as a local-first desktop evals, observability and AI memory project for Claude Code, Cursor, Windsurf, Cline, VS Code agents and custom SDK integrations; visible snippet listed fork count 2 and emphasized no Docker/server/cloud requirement. Shell GitHub API access remained blocked by DNS and local gh auth was invalid, so this card treats the current snapshot as web-observed rather than API-verified.

4. Teaching Use

Use this card to explain Local Agent Evals and Memory Observability in the raw -> classification -> project card -> site/report pipeline. Does observability plus local memory improve agent debugging and repeatability, or only collect traces?

5. Limits

The repository was not cloned in this iteration; no benchmark, SDK example, memory experiment, eval run, skill install flow, agent loop, or production deployment was executed. Counts and claims are visible public-page/search signals unless independently revalidated later.