exec reads now surface more of what OpenKnowledge already tracks, in…#426
Merged
Conversation
…) (#2345) * docs(open-knowledge): add Concepts section with interactive widgets New Concepts pages (What is an LLM wiki, Agentic search, Skills) with animated html-preview widgets that render on the docs site and in the OK editor. Adds a remark plugin turning html-preview fences into a sandboxed HtmlPreview iframe, wired into the nav. * feat(open-knowledge): exec surfaces templates and link sources in text (PRD-7251) The ls/cat enriched text now renders templates_available (name, description, scope), backlink and forward-link source paths on a single cat, the most-recent timestamp, and the direct md-file count. These rode on the structured payload but were omitted from the human-readable listing an agent reads. Covers PRD-7251 A1-A4; B1 (frontmatter in slim) and B2 (per-file mtime) remain. * feat(open-knowledge): exec surfaces graph role + status/type on reads (PRD-7251 Phase 1) computeGraphRole classifies a doc (hub/connector/leaf/orphan) from its in/out link counts and renders as `graph: <role>` on a single cat. formatFileEntry also renders triage-relevant frontmatter (status/type), which the enrichment already returns. Unit tests for the role thresholds; exec suite green. * docs(open-knowledge): agentic-search raw-vs-exec toggles + write panel + graph role Interactive cat/ls/grep/write html-preview widgets showing what exec/write return beyond raw shell; cat now shows graph role, grep shows status per hit, matching the shipped Phase 1 behavior. * docs(open-knowledge): spec + explainer for connectivity-native exec reads SPEC + SPEC-EXPLAINED for graph-native enriched reads and write advisories. Phase 0 (A1-A4) and Phase 1 (graph role, status/type) shipped; suggest flywheel split to PRD-7254; co-citation dropped. * fix(open-knowledge): drop useless escape in html-preview srcDoc (biome) * fix(open-knowledge): drop deprecated iframe scrolling attr in html-preview (oxlint) * fix(open-knowledge): address PR review — partial-data graph role, text-render tests, iframe clamp, changeset computeGraphRole returns null when either link count is unknown (no classify on partial data); text assertions for the ls count/recency rendering + partial-null unit tests; clamp html-preview iframe height; add changeset for the user-visible exec output change. * test(open-knowledge): pin graphRole/status/type render via exported formatFileEntry Exports the pure formatFileEntry formatter and unit-tests that graph role, status, and type appear in the text output, closing the review's render-coverage gap (graphRole needs a live server to exercise end-to-end, so the formatter is tested directly). * test(open-knowledge): cover backlink/forward-link source-path inlining in formatFileEntry * docs(open-knowledge): write widget renders row shows mermaid only (callouts have no render warning) * test(open-knowledge): pin graph-role boundary + formatFileEntry link edge cases * docs(open-knowledge): restructure concept docs + fix MCP page accuracy - Dissolve the top-level Concepts section; distribute its pages to where they belong: skills -> Features, agentic-search -> Reference (next to core-concepts), and merge the LLM-wiki idea into the karpathy-llm-wiki workflow (one page owns idea + build, no cross-section redirect). - Drop the read/write/idea triad framing between skills, wiki, and search. - Add interactive html-preview widgets to core-concepts (three layers, links/backlinks, link-graph tools, attribution) and link it to agentic search. - Correct the MCP reference against the actual tool surface: workflow gains the wiki kind; write/edit/delete/move gain the skill target; move has template+skill targets (not 'no template'); preview_url auto-starts the server rather than being server-free; history/restore_version/summary tool lists completed. * docs(open-knowledge): biome-format reference meta pages array * docs(open-knowledge): sync workflow-kind list + LLM-wiki link text after rename - Add the wiki kind to karpathy-llm-wiki's tool-kind enumeration; drop the bare 'three pipeline kinds' count in favor of naming the pipeline kinds so it can't go stale. - Update core-concepts cross-reference text to the renamed LLM wiki page. GitOrigin-RevId: 30f7541dc0827683e69dac2ab98cc16b10695582
![inkeep-internal-ci[bot]](https://avatars.githubusercontent.com/in/2806322?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Automated approval from agents-private public-mirror-sync (run: https://github.com/inkeep/agents-private/actions/runs/28549991358). Source of truth is the monorepo; direct edits on inkeep/open-knowledge are overwritten on next sync.
|
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
execreads now surface more of what OpenKnowledge already tracks, in the human-readable listing an agent reads.lsshows a folder's available templates (name, description, scope), its description and tags, and direct-vs-recursive file counts with the most-recent date;catshows the backlink and forward-link source paths (not just counts) and a coarse graph role (hub / connector / leaf / orphan); both render triage frontmatter (status,type) per file.