OpenClaw Enterprise Knowledge Base Skill

Skill X-RayLuan/awesome-ceo-stack/ceo-skill/OpenClaw-Enterprise-Knowledge-Base-Skill

An awesome, curated stack of Claude / OpenClaw agent skills for founders & CEOs — strategy, knowledge base, decks, K-12 coaching, and a full local AI promo-video pipeline. Works with Claude Code, Codex & OpenClaw.

Install

npx -y skills add X-RayLuan/awesome-ceo-stack --skill OpenClaw-Enterprise-Knowledge-Base-Skill

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
7 stars7 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when creating, auditing, repairing, validating, or syncing an enterprise Markdown or Obsidian knowledge base for OpenClaw agents. Applies to ontology-style wiki pages, governed business objects, hot.md startup cache, agent-safe visibility rules, OpenClaw manifests, RAG handoff, sales/support action governance, and knowledge-base QA workflows.

SKILL.md

4.9 KB, as published. Nobody here has run it

OpenClaw Enterprise Knowledge Base

Use this skill to turn enterprise wiki material into an agent-ready semantic layer for OpenClaw. The objective is not more notes. The objective is reliable agent behavior: verified facts, governed visibility, explicit relationships, approved actions, and a short startup cache.

Core Model

Treat every important Markdown page as a business object.

Markdown page = object
YAML frontmatter = object attributes
Wiki links / inline fields = relationships
Action pages = approved business verbs
status + visibility = governance and permissions
openclaw/manifest.json = runtime contract
openclaw/Corpus-Descriptions.md = retrieval routing map
openclaw/Agentic-RAG-Workflow.md = sufficient-context workflow
hot.md = fresh-session startup cache

When Starting

Decide whether the user is creating a new vault, auditing an existing vault, adding objects, or preparing OpenClaw sync.
For folder layout, read references/vault-structure.md.
For object fields and relationship style, read references/ontology-schema.md.
For object types, relationship predicates, action families, and governance vocabulary, read references/ontology-vocabulary.md.
For the root hot cache pattern, read references/hot-cache.md.
For OpenClaw runtime handoff, read references/agent-sync.md.
For agentic RAG, cross-corpus routing, and sufficient-context checks, read references/agentic-rag.md.

Ask only when the missing choice changes the artifact materially. Otherwise choose conservative enterprise defaults.

Required Object Frontmatter

Every object page must include:

---
type:
object_id:
status: needs_review
visibility: internal
owner:
source:
last_reviewed:
---

Allowed status values:

verified: may be used externally if visibility allows it.
needs_review: internal use only until a human verifies it.
conflicting: known conflict exists; do not cite externally.
deprecated: keep for history; do not use in active answers.

Allowed visibility values:

public: customer-facing.
dealer: partner-facing.
internal: staff-facing.
management: management-only.

Hot Cache Rule

Create a root hot.md for the current working context. OpenClaw should read it before Home.md.

Use hot.md for:

Current topic or active workstream.
Key recent conclusions.
Related pages to read next.
One concrete next step.

Do not use hot.md for:

Long-term policy.
Full knowledge dumps.
Customer-facing facts without source pages.
Secrets, credentials, private customer data, or sensitive pricing.

Action Governance

Actions are ontology verbs. Create action pages for agent behaviors such as quote drafting, product recommendation, service intake, or escalation.

Any action involving price, delivery date, warranty, contract terms, customer data export, public publishing, compliance claims, or deleting knowledge requires human approval.

Use the action families and approval gates in references/ontology-vocabulary.md when designing a reusable enterprise vault.

OpenClaw Startup Order

OpenClaw-facing vaults should include:

openclaw/manifest.json
openclaw/Corpus-Descriptions.md
openclaw/Agentic-RAG-Workflow.md
hot.md
Home.md

The manifest should expose safe paths, approved action names, startup order, retrieval routing files, and governance rules. It must not bypass status or visibility.

Agentic RAG And Sufficient Context

For vaults that support recommendations, quotes, support, compliance, or multi-source answers, add the cross-corpus pattern in references/agentic-rag.md.

Minimum behavior:

Classify the request workflow.
Route to the smallest sufficient set of corpora.
Retrieve frontmatter, direct evidence, and ## Relationships.
Check whether context is sufficient and export-safe.
If insufficient, retrieve again or ask the shortest missing follow-up questions.
Separate confirmed facts, drafts, missing fields, approval gates, and next actions.

Validation

Run the bundled validator from this repo:

node scripts/validate-vault.mjs --vault /path/to/vault

Before finalizing a vault, confirm:

Required frontmatter fields exist.
object_id values are unique.
Action pages include action_id.
Public pages are verified.
Action pages say what the agent may and may not do.
hot.md is short and contains no sensitive data.
OpenClaw startup order is explicit.