Guides the AIPCC package onboarding workflow: onboarding-cli triggers pipelines that open MRs (Builder failure path, RHAI Pipeline always, Probe Test failure path), merge order Builder then Pipeline then Probe, and Jira Epic/Stories with labels and transitions. Use when onboarding a new package, tracking onboarding MRs, or reconciling Jira with GitLab onboarding automation.
Before onboarding a package, load repo-specific and shared skills:
ai-helpers (always available):
${SAGENT_WORKSPACE}/ai-helpers/helpers/skills/python-packaging-complexity/SKILL.md
-- assess build complexity from PyPI metadata and suggest strategies.${SAGENT_WORKSPACE}/ai-helpers/helpers/skills/python-packaging-source-finder/SKILL.md
-- locate source repos for Python packages.${SAGENT_WORKSPACE}/ai-helpers/helpers/skills/python-packaging-license-checker/SKILL.md
-- check license compatibility with redistribution.${SAGENT_WORKSPACE}/ai-helpers/helpers/skills/python-packaging-env-finder/SKILL.md
-- discover build-time environment variables from setup/CMake configs.${SAGENT_WORKSPACE}/ai-helpers/helpers/skills/python-packaging-bug-finder/SKILL.md
-- find known packaging bugs, fixes, and workarounds via GitHub issues.Repo-specific skills:
${SAGENT_WORKSPACE}/builder/.claude/skills/package-settings/SKILL.md
-- generate package settings for the AIPCC wheels builder.${SAGENT_WORKSPACE}/builder/.claude/skills/prebuilt-package/SKILL.md
-- configure pre-built wheels that cannot be built from source.${SAGENT_WORKSPACE}/builder/AGENTS.md
-- builder repo conventions and plugin system.${SAGENT_WORKSPACE}/wheels-test/.claude/skills/probe-test-creator/SKILL.md
-- create probe tests for new packages in wheels-test.${SAGENT_WORKSPACE}/selfservice/.claude/skills/testing-builds/SKILL.md
-- test builds via self-service CLI and GitLab pipelines.Load all available skills from the lists above, then follow their instructions for the relevant phase of onboarding. Supplement with the generic steps below for workflow orchestration and anything not covered.
Automation flow
onboarding-cli kicks off the pipeline that creates or updates merge requests. Do not assume manual MR creation is the source of truth unless the user says so.
MR types
Merge ordering (blocking)
Merge in this order:
When using GitLab MCP or UI, respect MR dependencies / description notes so reviewers see the chain.
Jira structure
Operational steps
Use shell/ssh for CLI (onboarding-cli, git status on the remote server). Use GitLab MCP to list MR state, pipelines, and links. Use Jira MCP to create/update Epic/Stories, set labels, and transition issues.
Status reporting
Summarize: which MRs exist, merge order satisfied or not, open blockers, and Jira keys with current state.