rag-tools
Skill24 starsChanged 2 months agoMITView on GitHub
Rules for using the local-rag MCP server tools. Triggers when calling mcp_search_context, mcp_memorize, mcp_ingest_file, mcp_forget, mcp_list_topics, or any RAG-related operation.
In this repository43 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- anti-spiralSkill
- auto-orchestrateSkill
- b-plan-runnerSkill
- cookSkill
- dashboard-handoffSkill
- grillSkill
- idea-intakeSkill
- impact-checkSkill
- improvement-conjectureSkill
- incremental-evalSkill
- operatorSkill
- oracleSkill
- orchestrateSkill
- paper-to-planSkill
- pdfSkill
- planning-workflowSkill
- prompt-transformSkill
- seeSkill
- sequential-opus-orchestratorSkill
- spec-evalSkill
- teachSkill
- anti-spiralSkill
- auto-orchestrateSkill
- b-plan-runnerSkill
- cookSkill
- dashboard-handoffSkill
- grillSkill
- idea-intakeSkill
- impact-checkSkill
- improvement-conjectureSkill
- incremental-evalSkill
- operatorSkill
- oracleSkill
- orchestrateSkill
- paper-to-planSkill
- pdfSkill
- planning-workflowSkill
- prompt-transformSkill
- seeSkill
- sequential-opus-orchestratorSkill
- teachSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 29
If the MCP server is not running or returns an error, continue without it and note the failure in the response.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
RAG (local-rag MCP server)
You have access to a local offline knowledge base via MCP tools. Use them autonomously without asking the user.
All tools accept an optional project parameter for filtering (e.g. project="death-must-die"). Always pass the current project name when calling any RAG tool so results stay scoped. When ingesting or memorizing, always tag with the project name. When searching without a specific project need, omit the parameter to search across all projects.
mcp_search_context: call this during the search phase of every plan, before forming any hypothesis about a bug, and when entering unfamiliar code. Query with the concept, not the symptom. "velocity rotation hook pattern" not "crash".
mcp_memorize: call this after every completed plan phase and after fixing any bug. topic = short slug, tags = comma-separated categories. Do not wait until end of session. NEVER batch mcp_memorize with other tool calls. It triggers lazy model loading (280MB embedding model, 15-30 seconds on first call per session). Batching it blocks all queued tools and makes the UI appear frozen. Also: Qdrant uses an exclusive file lock -- never start a second Python process that imports the server while the MCP server is already running; it will deadlock with AlreadyLocked.
mcp_ingest_file: call this when a file is too large to read directly. Ingest it once, then query with mcp_search_context instead of reading it again.
mcp_forget: call this when a stored entry is confirmed wrong or outdated.
mcp_list_topics: call this at session start if the project uses the RAG heavily, to know what is already indexed.
Just-In-Time Verification Protocol: When retrieving archival memory or architectural rules via mcp_search_context, do not trust the memory blindly. Cross-reference stored citations against the live codebase. If the cited code has been altered and now contradicts the memory, or if the citations point to files that no longer exist, investigate the discrepancy and call mcp_memorize to update the archival memory before proceeding.
If the MCP server is not running or returns an error, continue without it and note the failure in the response.
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.