Extract structured digests from Claude Code and Codex session transcripts.
npx skills add https://github.com/ylxmf2005/AgentCorp --skill replay-ylxmf2005
Please help me install this Agent Skill. Skill: replay Source: https://github.com/ylxmf2005/AgentCorp/tree/main/agentcorp/replay Command: npx skills add https://github.com/ylxmf2005/AgentCorp --skill replay-ylxmf2005
Reconstruct and analyze how a work session actually unfolded from runtime transcripts to reveal where time and tokens were spent, what repeatedly failed, and how to turn those insights into actionable improvements for skills, docs, and conventions.
Run the extractor on your session transcript to generate a replay digest and an actionable report.
High-intent search queries and answers about installing and using this skill.
Analyzing session transcripts to find wasted tokens involves extracting a structured digest of turns, token counts, and tool interactions to reveal friction points. This identifies error bursts and long idle periods to pinpoint workflow inefficiencies.
Session trajectory analysis in Claude Code or Codex reconstructs how a work session unfolded from runtime transcripts to identify bottlenecks and repeated failures. It extracts metrics like token usage and tool calls to inform skill evolution and workflow conventions.
Generating an actionable report from Claude Code transcripts requires running the extractor on your session log to produce a replay digest. This digest routes findings to channels like skill-evolution, CLAUDE.md, or AGENTS.md for concrete follow-up actions.
Yes, session analysis improves documentation and conventions by identifying friction points and failure patterns in transcripts. Findings are routed to CLAUDE.md, AGENTS.md, or teamspace channels to generate concrete follow-up actions for workflow refinement.
Session replay works with both Claude Code and Codex transcripts. It extracts and assembles a structured digest from either platform's runtime logs, including turns, token usage, gaps, and tool interactions, to identify friction and opportunities for skill evolution.
Analyzing runtime transcripts for workflow improvements is limited to revealing friction within the recorded session, such as error bursts and idle periods. It cannot diagnose external system outages and requires routing findings to skill-evolution channels for manual workflow refinement.
markacianfraniExport active AI conversation transcripts into self-contained HTML files.
Arx-GameLog recurring tool-call failures and environment quirks into a persistent friction log.
ThedouglerTrack and analyze OpenRouter API usage, costs, tokens, and latency.
christophevgTranscribe conversations into structured transcripts with YAML frontmatter and synthesis sections.
