Claude Code against an Ollama-backed model via the headroom hub. Model
defaults to glm-5.3:cloud, overridable with OC_MODEL. Unproven here, so
route it at tasks whose failure is cheap to detect.
Ten git worktrees each carried their own 225 MB node_modules (1.1 GB across
five) and paid 11s per `npm ci`. pnpm hardlinks from a shared store: the same
five worktrees cost ~250 MB total, and a fresh install is 4s.
What changed beyond the mechanical rename:
- `overrides` moved to `pnpm-workspace.yaml`. pnpm 11 does not read the `pnpm`
field in package.json *or* npm's top-level `overrides`, and it fails silently
— the vite/defu/language-server pins would have quietly stopped applying.
- Build scripts are blocked by default in pnpm; esbuild and sharp are allowed
explicitly via `allowBuilds` (renamed from `onlyBuiltDependencies` in 11).
- `packageManager` + `engines` pin the toolchain.
- gate.sh rejects a package-lock.json/yarn.lock/bun.lock outright, so an agent
running `npm install` out of habit fails loudly instead of building a second,
divergent dependency tree.
- CI bootstraps pnpm with `npm install --global pnpm@11.25.0` rather than
corepack (unbundled as of Node 25) or pnpm/action-setup (this self-hosted
act-runner has never run a job; fetching a third-party action is not
something to discover on the first one).
Two pre-existing CI bugs fixed while in the file:
- the gate installed with `npm install --package-lock=false`, which discarded
the lockfile the previous session had just fixed.
- the visual-regression step imported `playwright`, which is not a dependency,
and `visual-regression.mjs` has no compare mode anyway — in CI it overwrote
its own baselines and passed unconditionally. Removed with a comment; it
comes back when it can diff.
The `publish` job is now manual (`workflow_dispatch`). During the migration
dist/ holds three HTML files against the live pages branch's ten, so publishing
on every push to main would take the site down to a stub. Restore at task 20.
HANDOVER.md's incident log still says npm where it describes what happened at
the time; that is history, not a missed rename.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The launcher read the agent out of each brief with [a-z-]*, which cannot match
content-i18n-migrator. Tasks 04, 05 and 06 silently fell back to
astro-architect. Now also fails loudly if the name is missing or has no agent
definition, rather than launching the wrong specialist.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The pre-existing root package.json has two scripts and no dependencies, and
there is no package-lock.json until task 01 scaffolds Astro, so every launch
died on `npm ci` before reaching the agent.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Routes each task to its CLI per MODEL-ROUTING.md (Codex for 01/03/15/16/19,
Gemini for 02/18, MiniMax for the rest), creates the worktree, installs the
toolchain once task 01 has produced a package.json, and runs the agent headless
with permission prompts disabled — these run unattended, and a blocked edit
would just hang.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Marcos Paulo