yiagent/OpenCI
AI-powered CI/CD toolkit — PR review, issue triage, smoke eval, observability, and more.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| task | Task to run. Supported values: pr/review — AI PR review (single PR comment) pr-agent/summarise — sticky CI summary on a PR pr-agent/feedback — @-mention agent on CI failure pr-agent/docubot — answer @docubot mentions pr-agent/test-gen — generate test scaffolds issue/orchestrate — issue agent action planning ci/smoke-eval — promptfoo / claude smoke eval stg/agent-test — autonomous L1–L4 staging tests prd-observe/canary-watch — 3σ post-deploy canary watch prd-observe/terraform-drift — daily Terraform drift check prd-observe/verify-fix — Sentry fix verification ops/error-triage — Sentry context for issue orchestration ops/summarize-failure — summarise a failed run ops/flag-audit — weekly feature-flag audit observability/health-report — daily AI-synthesised digest agents/ai-changelog — auto Keep-a-Changelog body agents/docubot — alias of pr-agent/docubot | yes | — |
| prompt | Custom prompt text or slash command. Overrides built-in skill prompt. | no | "" |
| prompt-path | Path to a custom prompt file (relative to caller repo). | no | "" |
| context | JSON object with scalar fields. Keys become {{name}} placeholders in prompt rendering. | no | {} |
| model | Claude model id. | no | claude-sonnet-4-5-20250929 |
| max-turns | Maximum agent turns. | no | 10 |
| anthropic-api-key | Anthropic API key. | yes | — |
| github-token | GitHub token for MCP tool access and CI data queries. | no | "" |
| api-base-url | Custom Anthropic-compatible base URL (e.g. self-hosted proxy). | no | "" |
Outputs
| name | description |
|---|---|
| execution-file | Path to the Claude Code execution output file. |
| branch-name | Branch created by Claude (when applicable). |
| session-id | Claude Code session id (usable with --resume). |