@inkeep/open-knowledge 0.87.3 → 0.87.5-beta.0

package/dist/THIRD_PARTY_NOTICES.md

@@ -6718,6 +6718,7 @@ The following MIT-licensed packages are patched in this repository via pnpm's `p
 | `cmdk@1.1.1` | `patches/cmdk@1.1.1.patch` |
 | `react-medium-image-zoom@5.4.3` | `patches/react-medium-image-zoom@5.4.3.patch` |
 | `react-resizable-panels@4.12.1` | `patches/react-resizable-panels@4.12.1.patch` |
+| `sirv@3.0.2` | `patches/sirv@3.0.2.patch` |
 | `y-indexeddb@9.0.12` | `patches/y-indexeddb@9.0.12.patch` |
 | `y-prosemirror@1.3.7` | `patches/y-prosemirror@1.3.7.patch` |
 

package/dist/assets/skills/project/SKILL.md

@@ -61,7 +61,7 @@ Tools NOT in OK MCP (your host's): `preview_start`, `preview_screenshot`, `WebFe
 - **Native `Read` / `Grep` / `Glob` on in-scope `.md` / `.mdx`** — the original case.
 - **`Bash ls` / `Bash find` / `Bash cat` on dirs containing in-scope markdown** — use `exec("ls -A …")` / `exec("find … -name '*.md'")` / `exec("cat …")`. Native returns bare names; `exec` adds frontmatter, backlinks, and recent activity. `-A` shows hidden entries without `.`/`..`.
 - **Glob patterns that target markdown** — `exec` expands file operands (`cat specs/*.md`); quoted patterns and a command's own pattern (`find -name`) stay literal.
-- **Dispatching the Explore / general-purpose subagent for markdown-heavy exploration** — subagents use native tools internally and bypass OK. Do markdown exploration yourself via `exec` / `search`. Subagents remain appropriate for **source-code** exploration.
+- **Dispatching the Explore / general-purpose subagent for markdown-heavy exploration** — a subagent may start without this skill loaded and then read markdown with its native tools, bypassing OK. Do markdown exploration yourself via `exec` / `search`. Subagents remain appropriate for **source-code** exploration.
 - **Native `Read` / `Grep` on in-scope markdown inside `.ok/`** — `.ok/` is in-scope; treat its `.md` / `.mdx` like any other KB file.
 - **`ls` / `cat` / `find` on skill folders to discover or read a skill** — skills are addressed by `name`+`scope`, not by path (a skill can live in any editor dir, the `.agents/skills/` hub, or a custom root, with copies elsewhere). Use the `skills` tool.
 

package/dist/assets/skills/project/references/conflict-resolution.md

@@ -28,7 +28,7 @@ A write that first detects a stale file during its final disk flush returns `urn
 | `working-tree` | a pull-only overlay collision, pinned to the origin blob it collided with | writes the chosen bytes to disk; nothing is committed |
 | `reconcile` | OK's own three-way merge of the editor against disk, including stale external saves. No git object holds the stages, so OK snapshots them at detection | writes the chosen bytes to disk and into the loaded document; no git command runs |
 
-A `reconcile` conflict carries a `reason`: `merged-with-markers` (the three-way merge produced markers), `refused-conflict-markers` (the disk bytes already carried markers), `refused-too-large` (the document is past the merge size cap), `disk-markers` (markers landed on disk under an open doc), or `stale-external-write` (an older displaced version returned on disk). The compatibility `conflictKind` field is `stale-external-write` for that last reason and `git` otherwise.
+A `reconcile` conflict carries a `reason`: `refused-conflict-markers` (the disk bytes already carried markers), `disk-markers` (markers landed on disk under an open doc), `refused-no-base` (there was no acknowledged version to merge against), `refused-too-large` (a merge past the size cap or one that failed OpenKnowledge's fidelity check, raised only when the outside version could not be saved to history), or `stale-external-write` (an older displaced version returned on disk). `merged-with-markers` appears only on entries recorded by earlier versions, or in the rare case where both versions were merged but could not be applied to the open document. An outside edit to the same block as edits made here is not a conflict: both versions are kept, this document's version first, and your write reports `disk-edit-reconciled` with `mergeOutcome: "both-kept"` (see `references/doc-editing.md`). The compatibility `conflictKind` field is `stale-external-write` for that last reason and `git` otherwise.
 
 ## What the user chooses
 

package/dist/assets/skills/project/references/doc-editing.md

@@ -10,4 +10,4 @@
 
 **MDX.** To author an MDX doc (the KB renders MDX/JSX components), set `extension: ".mdx"` on the create: `write({ document: { path: "guides/widget", content, extension: ".mdx", position: "replace" } })` lands `guides/widget.mdx`. A `.mdx` suffix typed into `path` works too (the `extension` field wins if you pass both); omit both and it lands `.md`. An existing doc keeps its on-disk extension regardless — changing it in place isn't available via the MCP today.
 
-**Advisory warnings on writes.** `write` and `edit` responses may include `structuredContent.document.warnings` (batch: per-doc `structuredContent.documents[].warnings`) — advisory entries discriminated by `kind`, each also summarized as a `⚠` line in the response text. The write always landed; the entries tell you what to do next. Write-integrity kinds mean re-read the doc (`exec("cat <path>")`) before continuing: `content-divergence` (`{ kind, intendedBytes, actualBytes, byteDelta, hint }` — the converged Y.Text doesn't match what the payload composed to: concurrent peer residue, or — rare — a primitive regression) and `disk-edit-reconciled` (an out-of-band disk edit was folded in before your write landed on top). The renderability kind `mermaid-parse-error` (`{ kind, fenceIndex, fenceFirstLine, message, line? }`) means that mermaid fence will not render — fix the fence and re-edit. The top-level `warning` field on write tools is unrelated: it is the preview-attach hint (`action: "attach-preview-once" | "start-ui"`), not an advisory — a separate key from `document.warnings`, and the two can coexist on one response.
+**Advisory warnings on writes.** `write` and `edit` responses may include `structuredContent.document.warnings` (batch: per-doc `structuredContent.documents[].warnings`) — advisory entries discriminated by `kind`, each also summarized as a `⚠` line in the response text. The write always landed; the entries tell you what to do next. Write-integrity kinds mean re-read the doc (`exec("cat <path>")`) before continuing: `content-divergence` (`{ kind, intendedBytes, actualBytes, byteDelta, hint }` — the converged Y.Text doesn't match what the payload composed to: concurrent peer residue, or — rare — a primitive regression) and `disk-edit-reconciled` (an out-of-band disk edit was folded in before your write landed on top; `mergeOutcome` says how: `clean`, `merged`, `both-kept` or `kept-live`). With `"both-kept"` the out-of-band edit changed the same blocks as edits made here, and three counts say what landed: `duplicatedBlocks` blocks are in the doc twice, this doc's version first and the out-of-band version directly under it; `restoredDeletedHere` blocks deleted here came back with the out-of-band version because it changed them; `restoredDeletedOutside` blocks the out-of-band edit deleted stayed with this doc's version because they were changed here. With `"kept-live"` it could not be merged, so the version here stayed and the out-of-band version was saved to the doc's history. Both carry `nextStep: "re-read"`, and in both your write landed: do not retry it. Re-read, and after `both-kept` remove a duplicate, or delete a block restored over a deletion here, with an ordinary `edit` if your task calls for it. A `mergeOutcome` or `nextStep` you do not know means the same as `merged`: re-read. The renderability kind `mermaid-parse-error` (`{ kind, fenceIndex, fenceFirstLine, message, line? }`) means that mermaid fence will not render — fix the fence and re-edit. The top-level `warning` field on write tools is unrelated: it is the preview-attach hint (`action: "attach-preview-once" | "start-ui"`), not an advisory — a separate key from `document.warnings`, and the two can coexist on one response.