miadi-orchestration-kit

OpenClaw Model-Routing Research Kit

Purpose-built Copilot plugin dir for the Miadi/OpenClaw model-routing study. It gives Iris/Hermes a reusable launch surface for study preflight, source ledgers, RISE rispec scaffolding, plugin-gap contracts, model-routing matrix work, and final integration handoffs.

Included Agents

Agent	Use
`openclaw-study-orchestrator`	Runs preflight, validates the launch manifest, enforces track order, scopes, token caps, and stop conditions.
`openclaw-source-ledger-reviewer`	Reviews claim ledgers, evidence quality, contradictions, and claim-to-source coverage before synthesis.
`openclaw-rispec-scaffold-reviewer`	Checks OpenClaw rispec folders against the RISE phase wording and the standard 5-file shape.
`openclaw-final-integrator`	Integrates track handoffs, source ledgers, routing matrix updates, plugin gaps, and acceptance status.

Included Skills

Skill	Use
`openclaw-model-routing-bootstrap`	Fixed-order study bootstrap, launch manifest validation, write-scope discipline, token caps, issue links, and stop rules.
`openclaw-source-ledger`	Source ledger rules for evidence types, quality ratings, contradiction handling, and claim-to-source mapping.
`openclaw-rise-rispec-scaffold`	RISE scaffold process using `Reverse-engineer -> Intent-extract -> Specify -> Export`.
`openclaw-plugin-gap-contracts`	Reconciles missing plugin names and prevents live connector gaps from blocking read-only research.
`model-routing-policy-simulator`	Produces or revises the model-routing matrix from task, risk, context, cost, latency, and review constraints.

Included Templates

Template	Use
`templates/launch-manifest.md`	Control file for track order, write paths, plugin dirs, add-dir paths, issue links, token caps, and stop conditions.
`templates/track-handoff.md`	Required closeout format for each delegated wave or Copilot track.
`templates/source-ledger.md`	Claim ledger with source type, evidence quality, contradictions, and verification status.
`templates/model-routing-matrix.md`	Seed model-routing matrix with required input and output columns.
`templates/acceptance-checklist.md`	Preflight and final acceptance checks for the study kit and track outputs.

Cheap Smoke Test

copilot --model gpt-5-mini --reasoning-effort high --yolo --no-ask-user \
  --plugin-dir /workspace/repos/jgwill/miadi-orchestration-kit/copilot/openclaw-model-routing-research-kit \
  --add-dir /workspace/repos/miadisabelle/workspace-openclaw \
  --add-dir /workspace/repos/jgwill/miadi-orchestration-kit \
  -p "List the loaded OpenClaw model-routing research kit agents and skills. Then identify the launch-manifest template path. Do not edit files."

Next Launch Shape

cd /workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE

copilot --yolo \
  --plugin-dir /workspace/repos/jgwill/miadi-orchestration-kit/copilot/openclaw-model-routing-research-kit \
  --plugin-dir /workspace/repos/jgwill/miadi-orchestration-kit/copilot/stckin-orchestration-kit \
  --add-dir /workspace/repos/miadisabelle/workspace-openclaw \
  --add-dir /workspace/repos/jgwill/miadi-orchestration-kit \
  --add-dir /workspace/repos/jgwill/miadi-orchestration-kit-storytelling-rispecs \
  --add-dir /workspace/repos/miadisabelle/mia-awesome-copilot \
  --add-dir /workspace/repos/jgwill/llms-txt \
  -p "@/deep-research
TODAY: 2026-05-08
Use the OpenClaw model-routing research kit. First validate launch-manifest.md, source-ledger template, and rispec scaffold status. Launch only the approved track named in the manifest. Respect token caps and write scopes."

Operating Scope

This kit is orchestration infrastructure. It does not implement live Discord, OpenClaw runtime, arXiv, web discovery, or GitHub code-search connectors. Those gaps are captured as contracts so the study can proceed with source ledgers and explicit mitigations.