Skip to content

Commit 224575d

Browse files
LCccodeclaude
andcommitted
security + docs: purge customer names + reposition as enterprise KB agent
- Remove real customer name "伊吾县教育局..." from skill examples (6 lines across skill.md + skill-claw.md); replace with generic "Acme Corp" placeholder. - Reframe README hero as enterprise-grade KB agent that mounts on OpenClaw / Claude Code as the long-term KB brain, with four explicit pillars: agent brain, cross-department co-creation, trainable KB engine, precision retrieval. Adds the mount-as-agent-brain callout under the hero. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
1 parent 8da1a14 commit 224575d

3 files changed

Lines changed: 17 additions & 8 deletions

File tree

README.md

Lines changed: 11 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -7,7 +7,9 @@
77
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](./LICENSE)
88
[![CI](https://github.com/LCccode/wikigraph/actions/workflows/ci.yml/badge.svg)](https://github.com/LCccode/wikigraph/actions/workflows/ci.yml)
99

10-
**The AI-native knowledge base builder, grounded in personal knowledge management theory, built for enterprise multi-department co-creation.** Type `/lcwiki` in Claude Code or OpenClaw to turn any folder of docs into a structured, collaborative, AI-queryable knowledge base — with **precision knowledge allocation** at ~10% the token cost of vanilla RAG and **~10× lower ongoing compile cost** via smart incremental ingest.
10+
**The enterprise-grade knowledge base agent, grounded in personal knowledge management theory, built for cross-department KB co-creation.** Type `/lcwiki` in Claude Code or OpenClaw — lcwiki plugs into your AI agent as its **long-term KB brain**, trains on any folder of docs your teams drop in, and answers cross-department queries with **precision knowledge allocation** at ~10% the token cost of vanilla RAG and **~10× lower ongoing training cost** via smart incremental ingest.
11+
12+
> **Mount lcwiki on OpenClaw or Claude Code as a drop-in agent brain.** One `pip install`, one `lcwiki install --platform claude|claw`, and your AI agent has a persistent, structured, collaboratively-trained knowledge base it can query accurately — across sessions, across departments, across months of KB drift.
1113
1214
Fully multimodal. Drop in `.docx`, `.pdf`, `.xlsx`, `.pptx`, markdown, images, audio, or video — lcwiki converts everything to markdown, extracts per-doc structure, concepts with cross-department family aliases, and an interactive knowledge graph in one shot. Then it lets your AI query the KB with a three-layer token-first fallback: scan 100-token tldrs → fall back to article body → only touch raw content as a last resort.
1315

@@ -19,7 +21,14 @@ lcwiki sits at the intersection of three traditions that rarely talk to each oth
1921
2. **Enterprise knowledge management** — Notion / Confluence / Microsoft SharePoint: multi-user, permissioned, structured workflows, but fundamentally authoring tools with weak retrieval and no AI-native query.
2022
3. **AI-driven RAG** — LangChain / LlamaIndex / vanilla embedding retrieval: query-side intelligence, but no KB discipline — chunks, embeddings, and prayer.
2123

22-
**lcwiki takes the PKM foundation (theory that produces good KBs), scales it to enterprise multi-department co-creation (one inbox, many contributors, one canonical concept), and wraps it in AI-native precision allocation (the right piece of knowledge goes to the right question at the right cost).** Your sales team drops proposals, R&D drops specs, legal drops contracts, research drops papers — and a single shared concept index merges "Digital Platform" / "数字基座" / "Digital Base" into one canonical concept so cross-department queries don't miss connections.
24+
**lcwiki is the first enterprise KB solution that is simultaneously:**
25+
26+
- an **agent brain** — mount it on OpenClaw or Claude Code, and your AI has long-term, structured, queryable memory across sessions,
27+
- a **cross-department co-creation platform** — sales, R&D, legal, research all contribute to one shared KB through a single inbox; concept family aliases merge cross-department terminology automatically,
28+
- a **trainable KB engine** — you "train" lcwiki by dropping docs into `raw/inbox/` and running `compile`; no vector DB setup, no chunking strategy, no embedding model tuning — the LLM reads each doc once and writes the wiki,
29+
- a **precision retrieval system** — the three-layer query selects the smallest sufficient context per question, so the right knowledge reaches the right question at the right cost.
30+
31+
Your sales team drops proposals, R&D drops specs, legal drops contracts, research drops papers — and a single shared concept index merges "Digital Platform" / "数字基座" / "Digital Base" into one canonical concept so cross-department queries don't miss connections. The KB scales from personal (10 docs) to enterprise (1000+ docs across 20 departments) with the same architecture.
2332

2433
> **Inspired by Luhmann's Zettelkasten and Andrej Karpathy's modern `/raw` folder idea** — dropping every paper, screenshot, tweet, and whiteboard photo into one directory and expecting your AI to just *understand it all*. [safishamsi/graphify](https://github.com/safishamsi/graphify) turned that folder into a knowledge graph. **lcwiki takes it two layers further: a proper wiki *and* a graph *and* a precision-allocation query layer** — so your AI has long-term memory, a map, and a cost-aware retrieval strategy. Every doc gets a structured article with a 100-token tldr; every concept gets a standalone page with family aliases; every connection lives in a persistent queryable graph. All three layers are built in one shot by an LLM subagent pass, wrapped with CLI-atomic write-verify so agents can't silently corrupt the KB, and maintained long-term by a self-healing `/lcwiki audit` with LLM-as-judge checks.
2534

lcwiki/skill-claw.md

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1054,11 +1054,11 @@ After merging, rerun `/lcwiki graph` so the graph picks up the consolidated conc
10541054
### 使用
10551055

10561056
```
1057-
# 清掉"伊吾" 开头的旧记录(支持子串匹配)
1058-
/lcwiki update ~/.openclaw/lcwiki 伊吾县教育局智慧教育应用建设方案
1057+
# 清掉"Acme" 开头的旧记录(支持子串匹配)
1058+
/lcwiki update ~/.openclaw/lcwiki Acme-Corp-Q3-Proposal
10591059
10601060
# OpenClaw 下 kb_path 可省略(用默认 KB)
1061-
/lcwiki update 伊吾
1061+
/lcwiki update Acme
10621062
```
10631063

10641064
### Step 1 — 查找匹配记录

lcwiki/skill.md

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -897,11 +897,11 @@ After merging, rerun `/lcwiki graph` so the graph picks up the consolidated conc
897897
### 使用
898898

899899
```
900-
# 清掉"伊吾" 开头的旧记录(支持子串匹配)
901-
/lcwiki update ~/.openclaw/lcwiki 伊吾县教育局智慧教育应用建设方案
900+
# 清掉"Acme" 开头的旧记录(支持子串匹配)
901+
/lcwiki update ~/.openclaw/lcwiki Acme-Corp-Q3-Proposal
902902
903903
# OpenClaw 下 kb_path 可省略(用默认 KB)
904-
/lcwiki update 伊吾
904+
/lcwiki update Acme
905905
```
906906

907907
### Step 1 — 查找匹配记录

0 commit comments

Comments
 (0)