[fas89]

Description

This cuts the 0.29.0 release notes. ## [Unreleased] moves under ## [0.29.0] — 2026-11-12, with an empty [Unreleased] above it. The founder asked for this release in session chat on 2026-11-12. Once this merges and ci.yml passes on the merge commit, v0.29.0 gets tagged and release.yml publishes it.

release.yml's own awk extractor, run on [0.29.0], finds 22 entries under Added, Changed, Fixed and Security. Since v0.28.0 the release contains:

• fix(ai): real endpoints, --agent AI mode, claude auth errors, keyless auto-pick, an isolated claude -p #850: real endpoints, --agent AI mode, claude auth errors, keyless auto-pick, an isolated claude -p;
• feat(agent-skills): an Agent Skills pack, its Claude Code plugin, scaffold-ide install and an eval harness #851: the Agent Skills pack, its Claude Code plugin, the scaffold-ide install, and an eval harness;
• fix(ai): #850 follow-ups — target-dir event, one Ollama reader for any host, no gemini fallback #852: one Ollama reader for any host, the stated endpoint for model lookups, and no gemini fallback;
• security(ai): FLUID_LLM_API_KEY only for a named provider; saved keys stay off project hosts #853: FLUID_LLM_API_KEY goes only to a named provider, and saved keys stay off project hosts;
• fix(ai): saved AI choice in non-interactive runs, JSON-only --agent stdout, mission agent loop #854: the saved AI choice in non-interactive runs, JSON-only --agent stdout, and the mission agent loop;
• fix(local): grant each output's tmp_ sibling, so a re-run replaces it #844: a re-run replaces an embedded-SQL output outside the contract directory.

Two corrections make the notes accurate:

• fix(ai): saved AI choice in non-interactive runs, JSON-only --agent stdout, mission agent loop #854 merged without a CHANGELOG change, so its three entries are added here.
• fix(ai): #850 follow-ups — target-dir event, one Ollama reader for any host, no gemini fallback #852's no-provider entry quoted a "does not read the saved choice" message variant. fix(ai): saved AI choice in non-interactive runs, JSON-only --agent stdout, mission agent loop #854 replaced that variant, so the entry now names the saved coding agent the run passes over.

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would break existing functionality)
• Documentation update
• Refactor / code cleanup
• New provider or provider enhancement

Documentation

• Docs PR linked:
  • Docs PR:
• No docs needed — this change does not affect user-facing behaviour
  • Justification: a release cut, CHANGELOG only. The forge_docs PR for security(ai): FLUID_LLM_API_KEY only for a named provider; saved keys stay off project hosts #853 and fix(ai): saved AI choice in non-interactive runs, JSON-only --agent stdout, mission agent loop #854 is in progress separately.
• Docs TODO — I know docs are needed and will create a PR in forge_docs before this is merged

Checklist

• I have read the Contributing Guide
• My code follows the project's coding standards
• I have added tests that prove my fix/feature works
• All new and existing tests pass (pytest)
• I have run ruff check and black with no errors
• New maintained Python files include license headers (python scripts/add_license_headers.py; examples/** exempt)

Testing

• release.yml's extractor finds 22 entries under 4 headings, and [Unreleased] is empty.
• product_guardrail passes, both in self-test and on the canon.
• detect-secrets-hook on CHANGELOG.md exits 0.
• The tests that read the CHANGELOG pass: 45 passed, 4 skipped.
• An open Trello card mentions 0.29.0, the deptry card, but it is in Done, so nothing gates the cut.