WIP: skill readiness probes (readiness trait, MockProbe, doctor --json)

[CCChisato]

Refs #4407

[github-actions]

Thanks @CCChisato for taking the time to contribute.

This repository is observing a maintainer-managed PR intake gate in dry-run mode, so this pull request is staying open. This note helps maintainers prepare the allowlist before any enforcement is considered.

Please read CONTRIBUTING.md for the expected contribution shape. A maintainer can grant recurring PR access by commenting /lgtm on a pull request.

[Hmbown]

Thanks @CCChisato — the readiness boundary in this draft is relevant to #4407, and I am keeping it in the v0.9.1 intake rather than losing it.

Before this can run as a release candidate, please narrow the branch to the readiness implementation and its canonical tests/docs:

• remove issues.json, issues2.json, and the unrelated my-docs/** architecture dump from this PR
• replace the [skip ci] WIP commit with a normal DCO-signed update so the full matrix runs
• add focused coverage for ready, missing executable/runtime, and probe failure JSON states without starting or installing dependencies
• keep the report explicit about detected versus installed versus runnable; do not claim a managed dependency runtime until that backend exists
• mark the PR ready only after the focused doctor/runtime tests and cross-platform CI are green

The core trait/registry direction is worth continuing. Once the branch is narrowed, I will review it as a real v0.9.1 candidate and preserve your authorship.

[Hmbown]

v0.9.1 is now freezing without a branch update after the requested narrowing. I’m preserving this draft and your authorship intact, but deferring the PR with its owning #4407 work to the v0.9.3 runtime/provider/tool architecture lane. The readiness trait/registry direction remains welcome; the existing narrowing, focused readiness-state tests, normal DCO, and full-CI requirements still apply before review.

[Hmbown]

Thanks — the ReadinessProbe trait plus ProbeRegistry, the MockProbe/NeverReadyProbe test doubles with isolation tests, and the doctor --json per-skill readiness/required_tools output are the right shape for #4407. Status is unchanged from the July 17 notes: this rides with #4407 in the v0.9.3 lane, and the narrowing, DCO, and full-CI requirements still apply.

Concrete items for the next push, beyond those already listed:

1. The current head commits unresolved conflict markers (<<<<<<< Updated upstream / >>>>>>> Stashed changes) in skills.rs, system.rs, main.rs, runtime_api.rs, and slash_menu.rs — the branch cannot compile. It looks like a stash-pop that was committed mid-resolve.
2. Drop the debug leftovers: the eprintln! in skill_readiness_badge and the std::fs::write("D:\\codewhale_debug.txt", ...) in list_skills.
3. Give ProbeRegistry one owner. It is constructed empty at each call site (inspect_skills, list_skills, the slash menu), so probe() returns None everywhere except doctor. Populate it once via install_system_skills and share that instance — and never run python3 subprocess probes per keystroke in the slash menu; probe once and cache.
4. runtime_api::list_skills hardcodes Unknown; wire it or omit the field until it is real.
5. Per v0.9.1: Report artifact-skill readiness and define a managed dependency runtime #4407, PDF reading should report Ready via the bundled Rust extractor with no Python present; presentations/spreadsheets probes exist, pdf/documents are still missing.

Also rebase — main.rs and runtime_api.rs have churned upstream since your July 17 base.

---

Generated by Claude Code