wiki
Skill4 starsChanged 4 months agoMITView on GitHub
Markdown-first knowledge base where the LLM acts as librarian. Ingests raw sources, compiles and interlinks topic files, self-maintains an index. No vector DB or embeddings required -- uses LLM-native navigation over structured markdown up to ~400K words.
42 other projects use a version of this.
Adopted by
Most recently taken up by
- WALKERKILLER/Look-Tongji-Notestheir own version, first seen Nov 8, 2026
- Yeachan-Heo/oh-my-claudecodetheir own version, first seen Nov 8, 2026
- SethGammon/Citadelword for word, first seen Nov 4, 2026
- zlatko-lakisic/agent-cooptheir own version, first seen Nov 1, 2026
- Lixiuxiu559/AgentForgetheir own version, first seen Nov 1, 2026
- saksham10arora-dotcom/everything-claude-codextheir own version, first seen Nov 1, 2026
Used with
Projects with it also use ai-slop-cleaner, ralplan and 8 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
Jurel89Skillai-slop-cleaner
Clean AI-generated code slop with a regression-safe, deletion-first workflow and optional reviewer-only mode
Also in 4 projects that use this
ibahgatSkillralplan
Create a strict, quality-first execution plan with explicit gates and rollback-ready sequencing.
Also in 4 projects that use this
Yeachan-HeoSkillultragoal
Create and execute durable repo-native multi-goal plans over Codex goal mode artifacts.
Also in 3 projects that use this
woosikkimSkilldeepinit
Generate hierarchical AGENTS.md codebase maps for agent context
Also in 3 projects that use this
ibahgatSkillhud
Control omi visual HUD profile and keep status output readable in normal, compact, or hidden mode.
Also in 3 projects that use this
klauswgSkilldeep-interview
在明确批准执行之前,通过数学化歧义门控进行苏格拉底式深度访谈
Also in 4 projects that use this
Other versions42 projects use a version of this.
Word for word · 1
- SethGammon/Citadel★ 927100% same
Edited · 0
None found.
Wrote their own version · 41
- Yeachan-Heo/oh-my-claude-sisyphus★ 41ksame name
- Yeachan-Heo/oh-my-claudecode★ 41ksame name
- Yeachan-Heo/oh-my-codex★ 34ksame name
- AgriciDaniel/claude-obsidian★ 16ksame name
- saltmd/salt.md★ 178same name
- ekadetov/llm-wiki★ 108same name
- saksham10arora-dotcom/everything-claude-codex★ 74same name
- skillberry-ai/cap-evolve★ 69same name
- WALKERKILLER/Look-Tongji-Notes★ 53same name
- 6eanut/llm-wiki★ 43same name
31 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository50 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- architectSkill
- archonSkill
- ascii-diagramSkill
- autopilotSkill
- costSkill
- create-appSkill
- create-skillSkill
- daemonSkill
- dashboardSkill
- decision-mapSkill
- deploy-stewardSkill
- designSkill
- doc-genSkill
- doSkill
- evolveSkill
- experimentSkill
- fleetSkill
- grillSkill
- housecleanSkill
- improveSkill
- infra-auditSkill
- learnSkill
- live-previewSkill
- loopSkill
- mapSkill
- marshalSkill
- merge-reviewSkill
- organizeSkill
- postmortemSkill
- prdSkill
- pr-watchSkill
- qaSkill
- refactorSkill
- research-fleetSkill
- researchSkill
- reviewSkill
- scaffoldSkill
- scheduleSkill
- session-handoffSkill
- setupSkill
- systematic-debuggingSkill
- telemetrySkill
- test-genSkill
- triageSkill
- unharnessSkill
- verifySkill
- watchSkill
- workspaceSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
13 rules. 0 are common, 3 are distinctive, and 10 are very specific to this project.
0
Common
Most of the 35 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
10
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 57
Create wiki/, wiki/raw/, wiki/topics/, an empty wiki/index.md, and wiki/.wiki-meta.json with fields: created, lastUpdated, topicCount: 0, …
- Very specific: line 65
Step 2: Write raw content to wiki/raw/source-{timestamp}.md with header: title/URL, ingested date, type, original reference.
- Very specific: line 73
Step 6: Update wiki/.wiki-meta.json with new counts.
- Very specific: line 79
Read wiki/index.md, identify 1-5 relevant topic files, explain reasoning briefly
- Very specific: line 80
Read those topics; follow [[cross-links]] up to 2 hops if relevant
- Very specific: line 82
If the wiki lacks enough information: state what it knows, list gaps, suggest /wiki --add for missing areas
- Very specific: line 86
Read meta, count topic files and raw sources, check for orphaned topics, broken cross-links, and stale topics (30+ days).
- Very specific: line 90
Read wiki/index.md and all topic files
- Very specific: line 98
Read all topic files, extract title/summary/date, write a fresh wiki/index.md.
- Very specific: line 102
.planning/does not exist: Run /do setup first to initialize the harness state directory. - Distinctive: line 120
Every topic file must have a title, last-updated date, and sources list
- Distinctive: line 121
Every topic file must have at least one cross-link to another topic (unless it is the only topic)
- Distinctive: line 123
No duplicate topic files (same slug = same file)
“Common” is measured: kept by at least 6 in 10 of the 35 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
/wiki -- LLM-Native Knowledge Base
Orientation
Use when: building and querying a markdown-first knowledge base -- ingests raw notes, deduplicates, surfaces answers. Don't use when: capturing session learnings into the evolve pipeline (use /learn); generating structured code documentation (use /doc-gen).
Directory Structure
wiki/
index.md # Master index
raw/ # Unprocessed sources (timestamped files)
topics/ # Compiled topic files (one per topic, interlinked)
.wiki-meta.json # Stats: topic count, source count, last compactionCommands
| Command | Behavior |
/wiki | Status overview: topic count, last update, pending raw sources |
/wiki --add [source] | Ingest a new source into the wiki |
/wiki --query [question] | Answer a question using wiki knowledge |
/wiki --status | Detailed wiki health: topic count, staleness, orphan detection |
/wiki --compact | Merge, deduplicate, and reorganize topics |
/wiki --rebuild-index | Regenerate index.md from current topic files |
/wiki init [path] | Initialize a new wiki at the specified path |
Protocol
Command: /wiki init [path]
Create wiki/, wiki/raw/, wiki/topics/, an empty wiki/index.md, and wiki/.wiki-meta.json with fields: created, lastUpdated, topicCount: 0, sourceCount: 0, totalWords: 0, lastCompaction: null. Default path: wiki/ at project root.
Command: /wiki --add [source]
Ingest a new source into the wiki.
Step 1: Determine source type — URL (fetch with WebFetch), file path (read), raw text (use directly), no argument (ask user).
Step 2: Write raw content to wiki/raw/source-{timestamp}.md with header: title/URL, ingested date, type, original reference.
Step 3: Identify 1-5 topics. For each: check if a topic file already exists — append to existing or create wiki/topics/{slug}.md.
Step 4: Each topic file has: title, > Last updated, > Sources, compiled content, and ## Related Topics with [[slug]] cross-links. Scan existing topics for cross-links when writing.
Step 5: Regenerate wiki/index.md with stats header, Topics table ([[slug]] | summary | last updated), and Recent Sources table.
Step 6: Update wiki/.wiki-meta.json with new counts.
Step 7: Output: source description, topics created, topics updated, index counts.
Command: /wiki --query [question]
- Read
wiki/index.md, identify 1-5 relevant topic files, explain reasoning briefly - Read those topics; follow
[[cross-links]]up to 2 hops if relevant - Produce a clear answer citing specific topic files used
- If the wiki lacks enough information: state what it knows, list gaps, suggest
/wiki --addfor missing areas
Command: /wiki --status
Read meta, count topic files and raw sources, check for orphaned topics, broken cross-links, and stale topics (30+ days). Output: topic/source/word counts, last updated/compaction dates, and health issues with counts and lists.
Command: /wiki --compact
- Read
wiki/index.mdand all topic files - Identify: merge candidates (overlapping subjects), split candidates (multiple distinct subjects), stale/outdated content, duplicates
- For each change: describe what and why, make the change, update all affected cross-links
- Rebuild
wiki/index.mdand updatewiki/.wiki-meta.jsonwithlastCompactionand new counts - Output: topics merged/split/removed/resolved, final count, estimated words
Command: /wiki --rebuild-index
Read all topic files, extract title/summary/date, write a fresh wiki/index.md. Output: "Index rebuilt with {count} topics."
Fringe Cases
.planning/does not exist: Run/do setupfirst to initialize the harness state directory.- No wiki directory: Prompt to run
/wiki init. Do not auto-create on query or status. - Source >50K words: Split into sections, warn the user.
- Topic name collision: Merge into existing topic, do not overwrite.
- Empty wiki queried: "The wiki is empty. Add sources with
/wiki --add." - Broken cross-link: During compaction/rebuild — flag. During --add — create stub
[[missing-topic]] (stub -- needs content)if no context. - URL fetch fails: Report failure, suggest pasting content directly.
- Wiki >~400K words: Warn during --status, suggest archiving or
/wiki --compact.
Contextual Gates
Disclosure: "Updating wiki at .planning/wiki/. Files will be created or modified."
Reversibility: amber — creates and modifies .planning/wiki/ files; undo by deleting or reverting changed files.
Trust gates:
- Any: all wiki commands (init, add, query, status, compact, rebuild-index).
Quality Gates
- Every topic file must have a title, last-updated date, and sources list
- Every topic file must have at least one cross-link to another topic (unless it is the only topic)
- The index must accurately reflect all topic files (no orphans after --add or --compact)
- No duplicate topic files (same slug = same file)
- Raw sources are preserved in wiki/raw/ and never deleted
- The --query command must cite specific topic files, not fabricate information
Exit Protocol
Output a summary appropriate to the command executed, then:
---HANDOFF---
- Wiki: {command executed} at {wiki path}
- Topics: {count} total, {new/updated/merged count} changed
- Status: {healthy | needs compaction | has orphans/broken links}
---Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

