Skip to content

yeaight7/agent-powerups / plugins/codebase-intelligence/skills/local-rag-mcp

local-rag-mcp

Verified

Use when querying, ingesting, or maintaining a local RAG MCP corpus for semantic document retrieval with privacy controls.

FileKindSize
SKILL.mdskill md3.4 KB
references/rag-tool-model.mdmarkdown1.5 KB

references/rag-tool-model.md

View on GitHub
# Local RAG MCP Tool Model

## Tools

| Tool | Use |
| --- | --- |
| `status` | Check DB size, chunks, documents, config, embedding backend. |
| `list_files` | Inventory corpus before ingest/query decisions. |
| `ingest_file` | Add/update approved PDF, DOCX, TXT, MD file by absolute path. |
| `ingest_data` | Add approved string, HTML, or Markdown with source metadata. |
| `query_documents` | Search with exact terms plus semantic context. |
| `read_chunk_neighbors` | Expand a hit before/after when context is incomplete. |
| `delete_file` | Remove stale or wrongly ingested file/source. |

## Query Pattern

1. Start with exact identifiers from user request.
2. Add domain context, not synonyms only.
3. Use `limit=5` for precision, `10` for broad exploration.
4. Read neighbors for definitions, conclusions, examples, or partial snippets.
5. Cite file/source and chunk context in synthesis.

## Ingestion Gate

Do not ingest by default. First answer:

- Is corpus already present?
- Is content approved for storage?
- Are embeddings local or remote?
- Are secrets, `.env`, private keys, logs, or auth files excluded?
- Is re-ingestion needed or will query suffice?

## Failure Modes

| Symptom | Likely Cause | Fix |
| --- | --- | --- |
| High-scoring irrelevant hits | query too broad | add exact file/name/domain terms |
| Missing adjacent rationale | chunk boundary split | use neighbors |
| Stale answer | outdated corpus | status/list then re-ingest scoped source |
| Privacy risk | broad ingest | narrow approved files and exclude sensitive dirs |
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall