@markstanden/opencode-docs-plugin 0.10.0 → 0.13.0

package/src/SKILL.md

@@ -1,79 +1,79 @@
 # Docs
 
-Durable knowledge lives in **lakes**: **find** it (search, then read), or
-**record** one focused doc at the narrowest true scope. Call the `docs` tool;
-inside `execute` it is `tools.docs.docs_search(…)`, same verbs.
-
-## Classes
-
-| Class              | Where                       | Use it for                              | Writes        |
-| ------------------ | --------------------------- | --------------------------------------- | ------------- |
-| **Instructions**   | nested `README.md`          | build, run, test, deploy this code      | never         |
-| **RepoContext**    | `<repo>/docs`               | this repo's traps, decisions, landmines | yes           |
-| **MachineContext** | personal lake (`~/doclake`) | tooling and traps across your machines  | yes           |
-| **GlobalContext**  | organisation lake           | standards and integrations across repos | yes, reviewed |
-| **Wiki**           | human-readable wiki         | narrative for humans                    | read-only     |
-
-Record at the **narrowest scope where the whole claim is true**. Two scopes
-means two docs; one omnibus doc is out of place everywhere.
+Durable knowledge lives in **lakes**: **find** it, or **record** one focused
+doc at the narrowest justified scope. Call the `docs` tool; inside `execute`
+it is `tools.docs.docs_search(…)`, same verbs.
+
+## Scopes
+
+| Scope            | Config    | Where                       | Use it for                        | Writes        |
+| ---------------- | --------- | --------------------------- | --------------------------------- | ------------- |
+| **Repository**   | `repo`    | `<repo>/docs`               | this repo's traps and decisions   | yes           |
+| **Personal**     | `machine` | personal lake (`~/doclake`) | a person's devices and setup      | yes           |
+| **Organisation** | `global`  | organisation lake           | shared standards and integrations | yes, reviewed |
+| **Instructions** | implicit  | nested `README.md`          | build, run, test, deploy          | never         |
+| **Wiki**         | `wiki`    | human-readable wiki         | narrative for humans              | read-only     |
+
+Place by dependency: this repo's code or decisions → Repository; a person's
+device, account or setup (a named host counts) → Personal; an established
+shared rule → Organisation, boundary stated — reusability alone is not
+ownership; operational steps → the README workflow; human narrative → the
+wiki's owners. Split only for claims with different owners; unresolved
+ownership stays an explicit proposal, never a guessed destination.
+
+`help` lists each lake's scope and description; `add` writes the sole writable
+lake without `lake`, else refuses with candidates — pass `lake`.
 
 ## Find it
 
-A **blank `search`** lists recent docs, one line each; a **`search`** ranks
-them, falling back to `Closest matches` and matching terms as **substrings** —
-so a typo finds nothing. Terms shorter than `MIN_TERM_LENGTH` (3) are dropped
-when longer terms exist; otherwise all query terms are kept. `help` lists the
-lakes and ranking.
+A **blank `search`** lists recent docs; a **`search`** ranks them, matching
+terms as **substrings** (a typo finds nothing) and falling back to
+`Closest matches`. Terms shorter than `MIN_TERM_LENGTH` (3) are dropped
+when longer terms exist; otherwise all query terms are kept.
 
-Open the match with `read`: the root-relative path, or `lake:path` when several
-lakes exist. A README result is `instructions:path` and opens with the built-in
-`read`. Consult matching docs before a broad grep; a tie ranks repo,
-instructions, global, machine, wiki. Knowledge lakes exclude `plans/`,
-`requirements/` and `wiki/` at the root; wiki lakes exclude only
-dot-directories and `node_modules`.
+Open the match with `read` (root-relative, or `lake:path`); a README
+result is `instructions:path` for the built-in `read`. Consult
+matching docs before a broad grep; a tie ranks repo, instructions, global,
+machine, wiki. Knowledge lakes exclude `plans/`, `requirements/` and `wiki/`
+at the root; wiki lakes only dot-directories and `node_modules`.
 
 ## Record it
 
-Write to `<area>/<title>.md` under the right lake; `CONVENTION.md` is
-authoritative.
-
-1. **Read first, never duplicate.** Read the docs relevant to the task: the same
-   area, keyword hits on its entities, or links from touched docs.
-2. **Choose one class** from the table — the narrowest that holds the claim.
-3. **Reconcile, then add.** Edit before add; merge near-twins; split a bloated
-   doc per topic; after a split, merge or move, repoint inbound references and
-   delete the emptied doc — no superseded stubs.
+Record to `<area>/<title>.md` under the right lake (`CONVENTION.md` is
+authoritative):
 
-One artifact per topic and class, never one dump; the compaction nudge shapes
-`Record with docs add:` blocks for this.
+1. **Read first, never duplicate:** the same area, keyword hits on its
+   entities, or links from touched docs.
+2. **Choose one scope** from the table — the narrowest that holds the claim;
+   one artifact per topic and scope.
+3. **Reconcile, then add:** edit before add; merge near-twins; split per topic,
+   repoint references, delete the emptied doc — no superseded stubs.
 
-Record through the typed `add`: metadata (`Area`, `Keywords`, `Summary`) and a
-body with an `H1` and a claim section — Facts, Inferred, Tested, Failed approaches, Open questions. Omit `path` to create `<Area>/<slug>.md`; pass `path` to replace **wholesale**, so omitted body content is lost and the near-twin reconcile is skipped. Refused: blank metadata, traversal, a non-`.md` or symlinked target, no claim section, and — only when `path` is omitted — a same-area near-twin.
+Record through the typed `add`: metadata (`Area`, `Keywords`, `Summary`) plus a
+body with an `H1` and a claim section — Facts, Inferred, Tested, Failed approaches, Open questions. `mode:"create"` writes a new doc (never overwrites; near-twins refuse unless `allowNearTwinReason` explains the distinction). `mode:"replace"` wholesale-replaces the doc at `path` against your `expectedRevision` (omitted content is lost; a missing target or stale token refuses, never creates). Refused always: blank metadata, traversal, a non-`.md` or symlinked target, no claim section.
 
-A doc is current or deleted; no `Status` or `Supersedes`. A failed attempt is
-valid knowledge when it records what was tried, why it failed, what is true now,
-and what not to retry.
+A doc is current or deleted; a failed attempt is valid knowledge (what was
+tried, why it failed, what is true now). Knowledge reads and writes end with a
+`docs: revision` line: the replace token.
 
-Lead with the actionable fact, keeping the reason, limits and evidence that
-affect using it; drop chronology, repeated conclusions and empty sections. Keep
-the service names, errors, paths and commands someone will search for. A
+Lead with the actionable fact, its reason, limits and evidence; drop
+chronology and empty sections. Keep the
+service names, errors, paths and commands someone will search for; the
 `Summary` decides whether to open the file.
 
 Never record secrets, credentials, tokens or customer data — name the vault or
-variable, never the value — and no personal data in GlobalContext or Wiki.
+variable, never the value — and no personal data in Organisation or Wiki.
 
 ## Check it
 
 `check` audits the **writable** knowledge lakes: front-matter, an `H1`, a claim
 section, no hand-maintained `Date`, a distinctive identity, near-twins and broken
-relative links. Obsolete fields, long docs and missing roots are **notes**, not
-findings. Validation follows writability — read-only lakes, wiki lakes and
-READMEs are not validated, and skipped lakes are named in the report. A very
-large report is capped with an `and N more lines not shown` tail; narrow it with
-`lake`. `add` enforces the structural rules at the write boundary;
-cross-document audits stay with `check`. `maxDocLines` (default 200, `0`
-disables) is a soft size signal, not a refusal.
-
-Run the project's tests and typecheck before proposing a commit, fixing
-failures rather than papering over them. Prepare the commit and **wait for
-explicit approval** — recording findings is not commit approval. Never push.
+relative links. Obsolete fields, long docs (`maxDocLines`, default 200, `0`
+disables) and missing roots are **notes**, not findings. Read-only lakes, wiki
+lakes and READMEs are searched but not validated; skipped lakes are named, a
+very large report is capped with an `and N more lines not shown` tail. `add`
+enforces the structural rules at the write boundary; cross-document audits
+stay with `check`.
+
+Run the project's tests and typecheck before proposing a commit; prepare it
+and **wait for explicit approval** — recording is not approval. Never push.