List and inspect Rune workflow run artifacts for debugging and retry planning.
npx skills add https://github.com/vinhnxv/rune --skill runs
Please help me install this Agent Skill. Skill: runs Source: https://github.com/vinhnxv/rune/tree/main/plugins/rune/skills/runs Command: npx skills add https://github.com/vinhnxv/rune --skill runs
Rune's runs skill provides quick access to per-agent workflow artifacts produced by Rune, speeding debugging and retry planning.
List the most recent runs with /rune:runs list to begin inspecting artifacts.
High-intent search queries and answers about installing and using this skill.
You can inspect per-agent workflow artifacts to debug a crashed agent by listing recent runs and viewing specific agent inputs, outputs, and timing data from read-only files like meta.json and run-index.jsonl.
The best way to surface failed agents in a Rune workflow for retry is to compute timing statistics across recent runs, which identifies and highlights agents that have crashed or failed execution.
You view timing statistics across multiple agent runs by accessing read-only run artifacts, allowing you to aggregate execution data and assess overall workflow health and performance.
Yes, you can use this approach to check recent workflow run statuses by listing recent runs with their corresponding agent counts and statuses to quickly understand workflow health.
No, you do not need write access to modify run-index.jsonl when debugging agent runs because the process relies strictly on read-only access to workflow artifacts to ensure safe and structured outputs.
Validating agent names matters when inspecting workflow artifacts because it guards against invalid paths and ensures the system provides structured, safe outputs when accessing run data.
lucasolsRun strict TypeScript evals for LLM/agent workflows via CLI.
HmbownRestore crashed or deleted systems from backups with stepwise verification.
nsheapsTrigger Claude workflows in GitHub Actions and monitor their runs via CLI or API.
jolibrainInspect machine-learning training runs via the training-observe CLI.
