Install probe is skipped because daily sync hits the 20-minute timeout

[blackdm666]

Summary

Newly discovered plugins can enter the dshfind base catalog, but their install state is never probed because the daily workflow reaches its 20-minute job timeout during pnpm sync:db. The later pnpm probe:install step is skipped on every run.

Reproduction / evidence

Affected plugin: https://github.com/blackdm666/dsh-plugin-88api-image

The public API row exists and its GitHub metadata/category are current, but install facts remain untouched:

{
  "is_plugin": null,
  "install": {
    "cmd": null,
    "kind": null,
    "pkg_name": null,
    "npm_published": false,
    "probed_at": null
  }
}

Actual package state:

• npm: dsh-plugin-88api-image@0.1.2 (latest)
• GitHub Release: v0.1.2
• root package.json declares dsh.bundle.patch: ./cordis.patch.yml
• stable package, canonical repository backlink, no lifecycle scripts

Affected workflow runs:

• 2026-08-23: https://github.com/hikariming/dshfind/actions/runs/32614954789
• 2026-08-24: https://github.com/hikariming/dshfind/actions/runs/32685974955

Both spend the remaining job budget in Run pnpm sync:db, are cancelled at 20 minutes, and skip Run pnpm probe:install, pnpm gen:plugins, and snapshot commit.

Impact

• Newly catalogued plugins remain is_plugin: null indefinitely.
• npm/Release/Git install methods are not detected.
• The detail page says the repository installation method has not been probed.
• Scoring and generated snapshots cannot observe the verified distribution state.

Could the discovery sync and install probe be split into separate jobs/workflows, or the sync made incremental/bounded so probe:install can run? A targeted/manual install probe for the affected plugin would also unblock this entry after the pipeline fix.