# Internal Knowledge

> Discover and retrieve from knowledge bases the current user can access.

- Skill: `zgiai/internal-knowledge` (Agent Skill)
- Install (CLI): `npx skillmds@latest add zgiai/internal-knowledge`
- Raw SKILL.md: https://api.skillmd.com/api/skills/zgiai/internal-knowledge/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Integrations & APIs
- Author: zgiai (https://skillmd.com/u/zgiai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/zgiai/internal-knowledge

---


# Internal Knowledge Skill

Use this skill to answer internal assistant questions with context from knowledge bases the current user can access.

## Workflow

1. First call `list_accessible_knowledge_bases` with a short query derived from the user's topic, business domain, or explicitly named knowledge base.
2. Inspect `status`, `fallback_used`, and returned knowledge base names/descriptions before selecting datasets:
   - If `status` is `success`, select only the most relevant returned `dataset_id` values.
   - If `status` is `fallback`, treat the candidates as weak matches. Select a dataset only when its name or description is clearly related; otherwise refine the list query once or ask the user which knowledge base/domain to use.
   - If `status` is `no_results`, answer from conversation context if possible and say no relevant accessible knowledge base was found.
3. Call `retrieve_knowledge` with the selected `dataset_ids` and a concise query. Do not use dataset IDs that were not returned by the list tool.
4. Inspect `status`, `source_summary`, `context_blocks`, and scores before answering:
   - If retrieved blocks answer the user, answer from those blocks.
   - If the blocks are missing, weak, or off-topic, rewrite the query once using clearer entities, synonyms, or constraints and retry with the same relevant datasets.
   - After two retrieval attempts, if the relevant answer is still unclear, ask the user for clarification or say that no relevant accessible knowledge was found.
5. When the user asks for original wording, definitions, synopsis text, policy clauses, exact wording, or "what does it say", quote or closely excerpt the retrieved source text first and cite the source. Summarize only when the user asks for a summary or the original text is too long.
6. When retrieved context is used, cite source names from `source_summary` or `retriever_resources` when useful.
7. Never expose internal IDs such as dataset ID, document ID, or segment ID to the user.
8. Never invent a knowledge-base answer when no relevant accessible context was found.

## Tool Usage

`list_accessible_knowledge_bases` accepts:

- `query`: optional search text for narrowing candidate knowledge bases.
- `limit`: optional maximum number of candidates. Defaults to 20 and is capped at 100.

`retrieve_knowledge` accepts:

- `query`: the user question or refined search query.
- `dataset_ids`: required selected knowledge base IDs returned by the list tool.
- `top_k`: optional maximum retrieved chunks. Defaults to 5 and is capped at 20.
- `retrieval_mode`: optional `hybrid`, `vector`, or `graph`. Omit it for the default hybrid mode; use `graph` only for relationship or entity questions.