feat(maf): consolidate refresh stack on current main

[OgeonX-Ai]

Summary

Consolidate the surviving autogen refresh work onto current main without reintroducing the stale dependency snapshot from the older PR stack.

Changes

• add the structured fallback telemetry module, CLI size guards, and regression tests
• add the deterministic critic engine and CLI with targeted tests
• keep the workflow hardening updates by pinning third-party GitHub Actions to full commit SHAs
• refresh the README and wiki with current verification guidance, repo-local virtualenv guidance for PEP 668 environments, and the MAF workflow/UI direction
• preserve the compatible Agent Framework dependency stack already on main

Why

autogen still had several open PRs that became stale after main moved forward. Those branches carried useful code and docs, but they also carried an outdated dependency snapshot that was failing CI and conflicting with the current branch head. This PR replaces that stale stack with one verified branch built on current main.

Validation

• python -m pytest tests/test_contract_compatibility.py -q --tb=short -s
• python -m pytest tests/test_critic.py tests/test_provider_fallback_telemetry.py tests/test_worker_boundary.py -q --tb=short -s
• python -m compileall autogen_starter autogen_dashboard maf_starter main.py -q
• node --check autogen_dashboard/static/app.js
• verified imports for agent_framework, agent_framework_devui, and agent_framework_orchestrations inside the repo-local .venv

Risks

• this consolidates multiple previously separate PRs, so reviewers should focus first on maf_starter/provider_fallback.py, maf_starter/critic.py, and .github/workflows/
• the repo-wide default pytest -q path still is not a reliable gate in this environment due to a capture/collection issue, so validation is intentionally targeted
• the docs describe the MAF 1.0 UI/workflow direction, but the operator shell is still the current local dashboard until a first-party DevUI or AG-UI surface is wired in

Related

• supersedes the stale work carried in PRs feat(28-02): structured JSON failure telemetry + CLI fallback size guards #12, ci: pin third-party actions to commit SHAs and least-privilege permissions #13, feat(29-01): deterministic peer critic pattern-scan engine #14, docs: Phase 36 README + docs/wiki refresh #15, and fix(deps): restore compatible agent framework stack #16
• follows the already-merged coverage baseline from PR test(coverage): ratchet autogen branch coverage gate #11

Reviewer notes

Start with the runtime changes under maf_starter/, then the workflow hardening in .github/workflows/, then the refreshed docs.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4894ebf56e

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Restrict Python exception checks to Python files

When the critic is run on a docs-only diff that adds a fenced Python example containing except Exception:, this loop still sends that Markdown file through check_bare_except and produces a blocking finding even though no executable Python changed. Limit the exception/telemetry checks to Python source paths so documentation or wiki updates cannot fail the critic gate for illustrative snippets.

Useful? React with 👍 / 👎.