📝 Restore unique maintainer content as CONTRIBUTING.md

After dropping project-specs.md last commit, three categories of
content were lost that have no other home in the repo:

1. The script prefix convention with its 'pick the right prefix,
   don't invent one' rule
2. The feedback-tier system (table + 'why these splits?')
3. The publishing workflow (tagged-release flow)

These are maintainer/contributor-facing material, not user-facing.
The standard OSS location for this kind of doc is CONTRIBUTING.md,
which keeps it separate from README.md (user docs) so the two
don't drift. README.md gains a pointer at the bottom.

The content is condensed: no config dumps, no restatements of
package.json, no historical commit-message examples. Only the
rationale that isn't already in the actual config files.

CONTRIBUTING.md

@@ -0,0 +1,54 @@
+# Contributing
+
+This document is for maintainers and contributors working on the project itself. End-user documentation is in [README.md](./README.md).
+
+## Script prefix convention
+
+Script names in `package.json` use a prefix that signals _when_ the script is intended to run. A `<prefix>:<name>` script is implicitly aggregated by a `<prefix>` script (if one exists) and run by the corresponding lefthook hook or CI step. Picking the right prefix documents the script's intended lifecycle:
+
+- `check:*` — read-only verification. Aggregated by `npm run check`. Used in pre-commit hooks (on staged files) and CI's build job (on the whole project). Read-only; never modifies files.
+- `fix:*` — mutating counterpart of a `check:*` script. Aggregated by `npm run fix`. Use after `npm run check` to auto-resolve issues; the diff is the review surface.
+- `test:*` — test scripts. `npm run test` is the canonical entry point (`check:tsc` + unit tests); `test:unit` skips the typecheck for fast local iteration; `test:ci` adds c8 coverage and is the CI variant.
+- `publish:*` — runs only at publish time, in the CI `publish` job (immediately before `npm publish`). There is **no** local `npm run publish` script — publishing is CI-only by policy. The `publish:` prefix still documents intent: this script validates the _publishable artifact_ (e.g., `dist/`) rather than the source.
+
+A new script should pick the prefix that matches its lifecycle, not invent a new one. If no existing prefix fits, that's a signal the script doesn't belong in the standard pipeline.
+
+## Feedback tiers
+
+The tools are organized into a feedback ladder. Each tier catches different things at different costs; the rule of thumb is "earlier tiers fire more often, faster tiers catch less, slower tiers are more thorough":
+
+| Tier              | When       | What it runs                                                | Time  |
+| ----------------- | ---------- | ----------------------------------------------------------- | ----- |
+| Pre-commit (auto) | on stage   | tsc + oxlint + oxfmt + cspell (staged files only)           | ~1.3s |
+| Pre-push (auto)   | on push    | `npm test` (full tsc + unit tests)                          | ~3.5s |
+| `npm run check`   | manual     | Full project audit: all `check:*` scripts + knip + outdated | ~10s  |
+| `npm run fix`     | manual     | Auto-resolve fixable issues (lint, format)                  | ~3s   |
+| CI build (auto)   | on push/PR | `npm run check` + `npm run test:ci`                         | ~30s+ |
+| CI publish (auto) | on tag     | `publish:publint` + `publish:attw`, then `npm publish`      | ~10s  |
+
+### Why these splits?
+
+- **`check:tsc`, `check:oxlint`, `check:oxfmt`, `check:cspell`** are in pre-commit because they are fast (~0.2–0.5s each), fully offline, and naturally scope to staged files via the `LEFTHOOK_FILES` env var convention. They give instant feedback on what you typed.
+- **`test` (and the `tsc` it includes) is in pre-push** because it runs the whole test suite across the whole project. The pre-commit `LEFTHOOK_FILES` convention doesn't apply to the test runner, so pre-commit isn't the right home. Pre-push runs after all commits are made but before the push leaves the machine, catching regressions that span multiple commits.
+- **`check:knip` and `check:outdated` are NOT in pre-commit** — knip scans the whole project (~4s, would noticeably slow the hook), and `check-outdated` queries the npm registry (~6.5s, network-dependent, advisory not correctness). Both run in `npm run check` and CI; pre-commit stays fast and offline.
+- **`publish:publint` and `publish:attw` are NOT in `check`** — they belong to the `publish:` prefix because they validate the _publishable artifact_ (`dist/`) and require a fresh build. Running them on every commit would be wasteful. They run in the CI `publish` job immediately before `npm publish`.
+
+### Before pushing
+
+Run `npm run check` locally. If you only trust pre-commit + CI, know that anything `check:knip` or `check:outdated` would catch will be caught by CI on the PR before merge.
+
+## Publishing workflow
+
+Publishing is CI-only by policy. Local `npm publish` is not supported.
+
+1. Develop and merge PRs to `main`.
+2. CI runs `npm run check` + `npm run test:ci` on every push and PR — this is the authoritative gate.
+3. After all intended changes are on `main`, bump the version locally:
+    ```sh
+    npm version <patch|minor|major>
+    ```
+4. Push the tag to GitHub:
+    ```sh
+    git push --follow-tags origin main
+    ```
+5. The `publish` CI job runs on the tag: `build` → `publish:publint` → `publish:attw` → `npm publish --access public`. The publish-tier checks must pass before the artifact is published.