agentleFS
IgorGanapolskyIgorGanapolsky/mac-yolo-safeguards/.agents/skills/dual-model-agentic-graphrag/SKILL.md

dual-model-agentic-graphrag

Skill3 starsChanged 4 days agoMITView on GitHub

Operationalizes the 3-Pillar Dual-Model Mixture-of-Agents & Agentic GraphRAG Playbook (Gemini 1M Strategist + Claude 200k Tactical Executor Split Workflow, Loop Memory Model, Schema-Constrained & Graph-R1 Hypergraphs with Rethink traversal, and Bifrost/LiteLLM/Portkey Cross-Provider Parity Routing with Rigid Schema Validation & 3-Hop Token Cap).

The file

Files
SKILL.md

Dual-Model Mixture-of-Agents & Agentic GraphRAG Playbook

When to Use

Invoke this skill whenever:

  • Coordinating Gemini Series (4 Argon / 3.1 Pro, 1M shared token buffer) as The Strategist / Global Ingester with Claude Series (Opus 4.7 / 3.5 Sonnet, 200k+ high-density tokens) as The Tactical Executor via the Split Workflow pattern.
  • Recording multi-run developer states, commits, and agent findings into Volodymyr Pavlyshyn's Loop Memory Model (~/.mac-yolo-safeguards/graphrag/loop-memory-graph.json).
  • Running Agentic GraphRAG with Pattern B (Schema-Constrained Ontology) and Pattern C (Graph-R1 Hypergraphs) with Format + F1 reward scoring and iterative rethink states.
  • Enforcing Dual-LLM Gateway Governance (Bifrost / LiteLLM / Portkey profiles):
    1. Cross-Provider Parity Routing (never fail over to a secondary region of the same provider).
    2. Rigid Structural Schema Validation (unparseable JSON or missing schema fields trigger an immediate LOGICAL_TIMEOUT failover).
    3. Token-Throttling Guardrails (MAX_FALLBACK_HOPS = 3 and $10/mo token budget cap).

CLI Commands

# Run full 3-pillar self-test (JSON or human output)
node tools/agentic-graphrag-hypergraph-engine.js --self-test
node tools/agentic-graphrag-hypergraph-engine.js --self-test --json

# Run Agentic GraphRAG Search Tier with Graph-R1 Rethink expansion & cited context output
node tools/agentic-graphrag-hypergraph-engine.js --query "prevent BTM background notification and duplicate HUD windows"

# Generate a Strategist -> Executor Split-Workflow Handoff Packet
node tools/agentic-graphrag-hypergraph-engine.js --split-handoff "Refactor HUD telemetry and verify anti-regression gate"

# Run the behavioral test suite
node tests/test-agentic-graphrag-hypergraph-engine.js

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.