Skip to content

knowledge

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

knowledge

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
.claude-plugin
.claude-plugin
 
 
reference
reference
 
 
skills
skills
 
 
vendor
vendor
 
 
CHANGELOG.md
CHANGELOG.md
 
 
README.md
README.md
 
 

README.md

knowledge

A Claude Code plugin that ingests external knowledge into durable, synthesized artifacts. Its first shipped pipeline distills a technical book (PDF or EPUB) into concept-organized, author-attributed skill reference files; a re-runnable setup action settles where synthesized artifacts land in the consuming repo.

Skills

Skill Invoke What it does
book-distill /knowledge:book-distill Turns a technical book (PDF/EPUB) into concept-organized, author-attributed skill reference files through a structured, multi-session read-write pipeline, updating the target skill's routing table.
course-digest /knowledge:course-digest Extracts and synthesizes an online video course (Dometrain, Teachable), transcripts, frames, resources, companion code, into repo-applicable recommendations. Actions: full pipeline (default), extract, analyze, status, resume, continue.
docpage-digest /knowledge:docpage-digest Ingests a single online documentation page (docs-site URL) into a verified knowledge slice. Unaltered original, SOURCES.md inventory, per-section model-matched digests, dual verification (one cross-vendor verifier), and an interview-ready handoff artifact. Publisher profiles (fetch channel, applicability filter, doc queue) are separable context files; first profile: Anthropic docs.
map-corpus /knowledge:map-corpus Maps a multi-resource documentation corpus into a verified, classified, triaged slice before any digesting: bounded discovery, a user-approved link map classifying every discovered URL, deterministic node manifests over immutable snapshots, and a per-node relevance inventory whose evidence a script gate verifies, handing an approved queue to N runs of /knowledge:docpage-digest.
video-digest /knowledge:video-digest Watches a single public video from YouTube or X (Twitter), transcript + visual frames, harvests reference links, drives external research, and synthesizes a prioritized repo-applicability menu. Actions: watch, queue, transcript, resume.
setup /knowledge:setup check (default) verifies library_dir against the repository's artifact convention and probes the extraction prerequisites read-only; apply routes personal option changes through Claude Code's plugin configuration prompt; apply install-deps provisions the video pipelines' node dependencies and Chromium.

What book-distill produces

  • Concept-organized reference files (60-160 lines each), named by what they teach rather than by chapter number, with the author attributed in section headers.
  • Routing-table and quick-decision-guide updates to the target skill's SKILL.md, so the skill loads the right reference file for a given developer question at query time.
  • Multi-author merges, where two books cover the same concept, their content is consolidated into a shared file.

You name the target skill when you invoke /knowledge:book-distill, so output lands somewhere you chose (${CLAUDE_PROJECT_DIR}/.claude/skills/<target>/): either an existing skill it extends or a new one it creates. Cross-session state (the file plan, page map, and a checklist) persists under ${CLAUDE_PLUGIN_DATA}, which survives plugin updates.

Usage caution. Copyright

This plugin is a neutral tool; you own the rights decision for everything you distill with it. A condensed distillation of a copyrighted book is a derivative work (17 U.S.C. §§ 101, 106), the copyright holder's exclusive rights include preparing and distributing derivatives, so distilled outputs carry redistribution risk. Keeping a private distillation for your own study is a different act from publishing, committing, or sharing one; fair use is a defense raised after the fact, not a safe harbor you can assume in advance. Publish, commit, or redistribute a distilled output only once you have satisfied yourself that doing so is lawful for that book. This is a caution, not legal advice.

The distilled output is written into a skill that Claude later auto-loads as model context, so review it before you commit or share it: treat the source book as untrusted input and confirm the distillation reflects the book rather than any instructions injected through its text.

Revisit condition

video-digest ships as a skill inside this plugin rather than a standalone plugin because a separate plugin cannot reach this one's vendored video-digestion package. Plugins are isolated, with no sibling reach-outs. Extract a standalone youtube plugin once that package is independently distributable, or a consumer needs video without the rest of the knowledge stack.

Requirements

  • A PDF or EPUB you have the right to read. PDF works natively with Claude Code's Read tool; EPUB requires unzipping and text extraction first.
  • Bash + coreutils for the skills' inline mechanics (book-distill hashes its progress-file slug with sha256sum/shasum on every run). On native Windows, install Git for Windows so they run under Git Bash, which bundles sha256sum.
  • unzip on PATH for the EPUB branch, not bundled with Git Bash; install it or extract the EPUB with another archive tool first. PDF-only use does not need it.

Install

/plugin marketplace add melodic-software/claude-code-plugins
/plugin install knowledge@<marketplace>

Migrating from the standalone book-distill plugin? Nothing to do. The marketplace's renames map migrates book-distill@<marketplace> to knowledge@<marketplace> automatically on your next session; the skill is now invoked as /knowledge:book-distill.

One exception: an in-progress multi-session distillation stores its resume checklist under the plugin's ${CLAUDE_PLUGIN_DATA} directory, which is keyed by plugin id and is not migrated by renames (that map rewrites enabledPlugins and pluginConfigs, not plugin data). If you have a distillation in flight, copy your old book-distill plugin-data directory to the new knowledge one before resuming so the resume pointer survives.

Configuration

Personal options, prompted by Claude Code at enable time (all optional; zero-config defaults keep every pipeline working):

Option Type Default Purpose
library_dir directory . (repo root) Directory where the plugin's ingestion pipelines land synthesized artifacts; a relative value resolves against the project directory. Portable non-project roots: an absolute path, a leading ~ (home-relative), or an env-var reference ${NAME} / %NAME% (e.g. ${KNOWLEDGE_CORPUS_DIR}) so a machine-varying root never needs a literal machine path in the stored value. Expanded when a pipeline resolves the root (the video-digest launcher and the docpage-digest work root today), failing loud on an unset variable. book-distill is unaffected. It writes to the target skill you name at invocation. A working-notes or artifacts convention declared in your own project's CLAUDE.md or rules takes precedence.
yt_dlp_js_runtimes string node video-digest: JavaScript runtime yt-dlp uses for YouTube signature deciphering. Set to off to omit the flag entirely.
yt_dlp_cookies_file string (empty) video-digest: path to a Netscape cookies.txt for authenticated acquisition. Never commit cookie files.
yt_dlp_cookies_from_browser string (empty) video-digest: browser to pull YouTube cookies from (chrome, firefox, edge, …), forcing one instead of the automatic fallback. A cookies file wins over this.
max_concurrent_acquires number 1 video-digest: cap on concurrent yt-dlp acquisitions during a batch (1–3). Higher increases HTTP 429 risk.

book-distill itself writes to a target skill you name at invocation, so it needs no configuration to run; library_dir is the shared artifact-landing seam the plugin's ingestion pipelines resolve through. The video-digest acquisition options above tune yt-dlp authentication and throttling; course-platform credentials are intentionally not userConfig. They stay in shell env vars because a sensitive option persists as plaintext on Windows today.

Options reference

Generated from this plugin's .claude-plugin/plugin.json. Every option Claude Code will prompt for when the plugin is enabled, with the environment variable each hook reads it from.

Option Type Default Environment variable Description
library_dir directory "." CLAUDE_PLUGIN_OPTION_LIBRARY_DIR Directory where synthesized knowledge artifacts land. Default is the consuming repo root; a relative value is resolved against the project directory. Portable non-project roots: an absolute path, a leading ~ (home-relative), or an environment-variable reference ${NAME} / %NAME% (e.g. ${KNOWLEDGE_CORPUS_DIR}) so a machine-varying root never needs a literal machine path in this stored value. A working-notes or artifacts convention declared in your own project's CLAUDE.md or rules takes precedence.
yt_dlp_js_runtimes string "node" CLAUDE_PLUGIN_OPTION_YT_DLP_JS_RUNTIMES JavaScript runtime yt-dlp uses for YouTube signature deciphering. Default 'node'. Set to 'off' to omit the --js-runtimes flag entirely.
yt_dlp_cookies_file string (none) CLAUDE_PLUGIN_OPTION_YT_DLP_COOKIES_FILE Path to a Netscape-format cookies.txt for authenticated video acquisition (YouTube bot checks; the three login-required X cases). Empty by default (unauthenticated; YouTube adds an automatic browser-cookie fallback on a bot check, and X never iterates browser profiles). Never commit cookie files.
yt_dlp_cookies_from_browser string (none) CLAUDE_PLUGIN_OPTION_YT_DLP_COOKIES_FROM_BROWSER Browser to pull cookies from (e.g. chrome, firefox, edge), forcing one instead of the automatic platform-ordered fallback. YouTube only, since the X adapter is cookies-file-only. Empty by default. A cookies file, when set, wins over this.
max_concurrent_acquires number
min 1, max 3 		1 CLAUDE_PLUGIN_OPTION_MAX_CONCURRENT_ACQUIRES Cap on concurrent yt-dlp acquisition runs during a batch. Default 1; raising it increases HTTP 429 throttling risk.

How to set these

Three supported routes, in the order most people want them:

  1. Interactively. Claude Code prompts for declared options when you enable the plugin. To change them later: /plugin configure knowledge@<marketplace>.

  2. Headless. Repeat --config for each option. Replace <marketplace> with the marketplace you installed this plugin from:

    claude plugin install knowledge@<marketplace> -s <scope> --config library_dir=<value>

    The same command reconfigures a plugin that is already installed: it prints already installed and still writes the value. The short-circuit message is about the install, not the config write. Do not claude plugin uninstall to reconfigure: uninstalling drops this plugin's whole stored pluginConfigs entry, resetting every option in the table above to its default. -s defaults to user, so pass the scope claude plugin list reports for this plugin. The verified-version record lives in the plugin-reconfiguration convention.

    The value is stored immediately; the session you are in does not change. Hooks are handed their CLAUDE_PLUGIN_OPTION_* when the session starts, so start a fresh Claude Code session before expecting new behavior. A check run in the old session still reports the old value, and that is not a failed write.

  3. By hand, in settings. Add the value under pluginConfigs in your user settings (~/.claude/settings.json):

    {
  "pluginConfigs": {
    "knowledge@<marketplace>": {
      "options": {
        "library_dir": <value>
      }
    }
  }
}

    Plugin option values are read from user, --settings, and managed settings only, not from a project's .claude/settings.json. To vary behavior per repository, enable or disable the plugin in that project's enabledPlugins instead of setting an option there.

Do not set the CLAUDE_PLUGIN_OPTION_* variables yourself. They are how Claude Code hands a configured value to a hook process; the value comes from the routes above.

Upstream documentation

License

MIT (SPDX-License-Identifier: MIT).