Cluster friction payloads and open descriptive GitHub issues per cluster.
npx skills add https://github.com/hcross/crewrig --skill harness-curator
Please help me install this Agent Skill. Skill: harness-curator Source: https://github.com/hcross/crewrig/tree/main/.gemini/skills/harness-curator Command: npx skills add https://github.com/hcross/crewrig --skill harness-curator
Harness friction signals accumulate across teams, and manual triage is slow and error-prone. This Skill automates ingestion of friction payloads from the harness-friction wing, clusters them into actionable groups, and surfaces a descriptive GitHub issue per cluster on the appropriate repositories, enabling faster feedback loops and measurable improvement.
Run the harness-curator to generate friction clusters and create GitHub issues from the friction wing.
High-intent search queries and answers about installing and using this skill.
To automate GitHub issue creation from friction payloads, you can use a curation workflow that clusters signals by subcategory and opens labeled issues on target repositories. This provides traceability by writing opened URLs back to your metadata drawers.
Clustering friction signals by severity improves triage by grouping related payloads into actionable buckets, labeling each GitHub issue with severity and dominant room metadata, and preventing duplicate tickets during scheduled maintenance sweeps.
Yes, you can scan transcripts for potential friction signals automatically using a deep mode. This mode analyzes transcript data to detect frictions and supports an auto mode with deduplication and maximum-issues controls for scheduled maintenance.
Yes, automated friction curation requires the GitHub CLI and the MemPalace Python library. These dependencies validate inputs, manage repository authentication, and write opened issue URLs back to the friction drawers for traceability.
The best way to deduplicate friction issues during scheduled sweeps is using an auto mode with built-in dedup controls and a maximum-issues cap. This validates inputs, prevents redundant tickets, and maintains end-to-end traceability through drawer metadata.
Automated friction issues may miss severity labels if input validation fails or payload metadata is incomplete. The curation process assigns labels like severity and room based on the worst severity and dominant room within each clustered friction group.
crewrigCluster structured harness friction reports into labeled GitHub issues.
zhouboyu-xrealSimulate a difficult non-technical user to identify UX issues and create tickets.
baekenoughMonitor hada.io RSS feeds and generate labeled GitHub issues for AI harness content.
tizeeCreate and run standalone Python scripts with inline dependency metadata using uv.
