Skip to content

docs: OpenCode template fix, How it works diagram and explainer video - #84

Merged
Telschow merged 3 commits into
mainfrom
claude/audit-repair-ai-projects-0z7j93
Oct 12, 2026
Merged

Telschow merged 3 commits into
mainfrom
claude/audit-repair-ai-projects-0z7j93

Conversation

@Telschow

Copy link
Copy Markdown
Owner

Context

An audit of the OpenCode/FreeLLMAPI setup, and a request for management-ready project material.

Change

  • opencode.example.json: a "disabled": true key inside a provider entry is not part of OpenCode's schema and is dropped silently, so Ollama stayed on. Replaced with disabled_providers. Checked against OpenCode 1.19.0 with a local OpenAI-compatible mock: @ai-sdk/openai-compatible works unchanged. New tests/test_opencode_example.py; a note in CONTRIBUTING.
  • README: a management-level "How it works" diagram (sensors, fault detection, filter, accuracy and honesty checks, gate B1), included on the project page.
  • Project page: embeds a captioned explainer (docs/assets/video/) rendered from one benchmark run (outage_control, outage_visual). No published number moves; the README gains no numbers.

How tested

  • pytest tests/test_opencode_example.py fails on the old template, passes on the new one.
  • navkit run --only outage_control outage_visual reproduced the README table exactly (3.760 m / 2.322 m, NEES 4.1 / 286.2, coverage 100.0% / 20.5%); the video reads its figures from that record.
  • Doc tests (test_check_links, test_doc_tables, test_quoted_metrics, test_hero) 54 passed; scripts/check_links.py 0 broken; scripts/scan_secrets.py clean; mkdocs build --strict clean; ruff clean.

Risks

The full suite did not finish in the sandbox (stopped at 15 minutes); CI runs it. The video adds about 1.3 MB to the repository.

Checklist

  • ruff, ruff format, targeted tests (full suite left to CI, see Risks)
  • Every number I quote in a doc comes from a command I ran; scripts/check_doc_tables.py (via test_doc_tables) still passes
  • No # noqa, no new mypy error-code suppression, no skipped or weakened test
  • No change to the filter or the fixture
  • No credential, key or personal data in the diff
  • Conventional Commit title; one logical change (two related doc changes in one PR)

🤖 Generated with Claude Code

https://claude.ai/code/session_017LmqBGdTpNrQBMr44osvmo


Generated by Claude Code

OpenCode 1.19.0 drops a 'disabled' key inside a provider entry without a
warning, so the Ollama provider stayed on. Checked against a local
OpenAI-compatible mock endpoint; @ai-sdk/openai-compatible works unchanged.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017LmqBGdTpNrQBMr44osvmo
A management-level diagram (sensors, fault detection, filter, the accuracy and
honesty checks, gate B1) goes in the README and, by include, on the project
page. The page embeds a captioned explainer rendered from one benchmark run
(outage_control and outage_visual); every figure in it is read from that
run's records, none is typed. No new numbers in the README.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017LmqBGdTpNrQBMr44osvmo
scripts/metrics.py --write after tests/test_opencode_example.py added three
tests (1319 to 1322); the lint job's metrics --check failed on the stale
count.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_017LmqBGdTpNrQBMr44osvmo
@Telschow
Telschow merged commit ffbcd7a into main Oct 12, 2026
19 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants