sk Skill · Continuum-AI-Corp
orca-replay
Answers questions about a past agent run from its recording rather than from memory, and replays or forks that run. Use when asked why an earlier run did something, or to reproduce a failure.
Open on skills.sh ↗read 2026-09-17
- installs 8w
- 0
- 30-day movement
- starts with the next reading
- Related entries
- 1
- Connections
- 0
bashTypeScript
orca-replay is a skill that answers questions about a past agent run from its recording instead of from memory, and can replay or fork that run. It relies on the `orcareplay` npm package (Node 20+) with its MCP server registered as `orca` and at least one recorded run in the project's .orca/runs directory.
Use it when you need evidence-based answers about why an earlier run did something, or want to reproduce a failure exactly as it happened.
Use it to
- Investigate why a run deleted, overwrote, or moved a file
- Identify which step broke the build
- Reproduce yesterday's failure via orca_replay
- Fork one run onto several models from the same checkpoint with orca_compare
- Export and scrub a run as a self-contained HTML file for sharing
For Developers debugging or auditing past AI agent runs
- Host repository
- Continuum-AI-Corp/OrcaReplay
- Version
- 0.1
- Compatible with
- Requires the `orcareplay` npm package (Node 20+) with its MCP server registered as `orca`, and at least one recorded run in the project's .orca/runs directory.
- Licence
- Apache-2.0
- Host stars
- 252
- Host language
- TypeScript
topicsagent-runsreplaydebuggingrecordingsreproducibilitymcp