Skip to content

Contributing

This accelerator is a Microsoft-maintained template for partners delivering agentic AI solutions to customers on Microsoft Agent Framework + Microsoft Foundry. Contributions — bug reports, doc fixes, pattern suggestions, new reference scenarios — are welcome from partners and the broader community.

Ways to contribute

  • File an issue for bugs, gaps, or questions: github.com/Azure-Samples/agentic-ai-solution-accelerator/issues
  • Open a pull request for small fixes (typos, broken links, clearer wording) directly against main.
  • Propose a new reference scenario or pattern via issue first so we can align on scope before you invest in the implementation.

Pull request basics

  1. Branch from main.
  2. Sign the Microsoft CLA if the portal indicates one is required — a CLA bot will comment on your first PR with a link. See .github/CLA.md for the pointer doc.
  3. Follow the relevant CODEOWNERS for required reviewers.
  4. All PRs run pre-merge validators (see .github/workflows/) — lint, strict MkDocs build, link check. Keep them green.
  5. Schema changes require schema-steward approval.
  6. RAI-relevant changes require RAI council review.

Change classes

  • Patch: bug fix, doc fix — any CODEOWNER + validators pass.
  • Minor: new T1/T2 behavior (backward-compat) — accelerator-eng approval.
  • Major / breaking: schema break, bundle change, new T2 package — accelerator-eng leads + RAI council review for RAI-touching changes.

Adding a reference scenario

  1. Propose via issue first — describe business problem, solution shape, topology, grounding model.
  2. Run python scripts/scaffold-scenario.py <slug> to materialize the scaffold under src/scenarios/<slug>/.
  3. Paste the printed scenario: snippet into accelerator.yaml (replacing the existing block on a reference branch, or use a variant file for multi-scenario demos).
  4. Land a README + walkthrough under docs/references/<slug>/ describing the business case, KPIs, and how the scenario was customized.
  5. Keep evals/quality/ and evals/redteam/ green; the lint (scripts/accelerator-lint.py) will enforce the manifest contract.

Adding a pattern doc

Pattern docs go in docs/patterns/<pillar>/. Keep them opinionated + short. Link to reference scenarios where they demonstrate the pattern.

Release process

Releases tag main when the flagship + framework + evals are green against the weekly pinned-latest CI. Release notes call out manifest changes (schema edits, new required keys) so partner forks can rebase cleanly.