@danypops/pi-pipes
Pi extension: agent-facing CI tool for cross-platform CI (GitHub Actions, GitLab CI, Jenkins)
Package details
Install @danypops/pi-pipes from npm and Pi will load the resources declared by the package manifest.
$ pi install npm:@danypops/pi-pipes- Package
@danypops/pi-pipes- Version
0.9.8- Published
- Aug 4, 2026
- Downloads
- 3,111/mo · 2,897/wk
- Author
- danypops
- License
- MIT
- Types
- extension
- Size
- 110 KB
- Dependencies
- 6 dependencies · 2 peers
Pi manifest JSON
{
"extensions": [
"./src/index.ts"
]
}Security note
Pi packages can execute code and influence agent behavior. Review the source before installing third-party packages.
README
@danypops/pi-pipes
The agent-facing side of pi-pipes: a single ci tool over
@danypops/pipes's operation registry — trigger, watch, and check
results for CI pipelines across GitHub Actions, GitLab CI, and Jenkins
without hand-rolling each backend's API. No network access or credentials
of its own; everything routes through the authenticated daemon client.
Bookmarked job templates (presets)
A preset is a name bound to a backend + an ordered list of steps (job name
- params), the same shape conty's own
pipelines.yamluses. Once bookmarked,trigger/status/logaddress it by name via thepipelineparameter instead of rawbackend+jobRef+paramsevery time.
The agent can manage these directly:
ci(action=presets)— list every bookmarked preset's full definition.ci(action=bookmark, pipeline=<name>, backend=<name>, presetSteps=[{jobName, params?}, ...])— save (or overwrite) one.ci(action=unbookmark, pipeline=<name>)— remove one.ci(action=trigger, pipeline=<name>, params={...})— trigger a bookmark with a per-invocation override, merged onto every step's own baked-in params (the override wins on key collision). Use this for a preset whose values legitimately change between runs (a release image, a branch) instead of re-bookmarking it just to update one value.
/pipes command
A human-driven TUI menu (backend/preset picker → job picker → action
picker) for manual trigger/cancel/log-viewing and preset management —
distinct from the agent-facing ci tool, no LLM involvement.