| name | openclaw-model-routing-bootstrap |
| description | Bootstrap the OpenClaw model-routing study by reading fixed-order context, validating or creating the launch manifest, enforcing write scopes, token caps, issue links, and stop conditions. |
Use this skill at the beginning of every OpenClaw model-routing Copilot session.
Fixed Reading Order
- Confirm the current date and timezone from the user prompt or environment.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/README.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/01-notions-map.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/02-mastery-path.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/04-build-ideas.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/preparing-deep-research/PROPOSAL.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/preparing-deep-research/KINSHIP.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/codex-delegation-readiness-review-2026-05-08/README.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/codex-delegation-readiness-review-2026-05-08/01-delegation-readiness-review.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/codex-delegation-readiness-review-2026-05-08/02-copilot-proposal-delta.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/codex-delegation-readiness-review-2026-05-08/04-plugin-dir-readiness-plan.md.
- Read
/workspace/repos/miadisabelle/workspace-openclaw/study-notes/openclaw-model-routing-85Q9htV2CBE/codex-delegation-readiness-review-2026-05-08/05-iris-orchestrator-needs.md.
- Read
/workspace/repos/jgwill/llms-txt/llms-rise-framework.txt if present.
If a required file is missing, record the missing path and stop before research unless the manifest already authorizes a fallback.
Manifest Process
- Locate
launch-manifest.md in the study root.
- If it does not exist and the user authorized preflight, create it from
templates/launch-manifest.md.
- Validate these fields:
- date and timezone, expected
2026-05-08 America/Toronto for the first launch
- orchestrator, expected
Iris/Hermes
- study root
- related issues, including
jgwill/miadi-orchestration-kit#14 and miadisabelle/workspace-openclaw#80
- plugin dirs and add-dir paths
- allowed write scopes
- track order and currently authorized track
- token caps
- one optional gap wave per track maximum
- stop conditions
- Do not launch a track when the manifest is absent, ambiguous, or contradicted by the user prompt.
Write Scopes
- Write only to paths declared in the manifest.
- For preflight, prefer the study root and explicitly named rispec folders.
- Do not edit connector repositories, plugin catalogs, or unrelated project files during read-only research.
- Treat live Discord and OpenClaw runtime connector work as out of scope unless the manifest explicitly changes.
Token Caps
- Preflight: <= 25k
- Academic: <= 230k
- Technical: <= 320k
- Narrative: <= 190k
Before launching work, estimate the track budget. Stop if the requested wave would exceed the cap or require a second gap wave.
Stop Conditions
Stop and report when:
- manifest is missing and creation was not authorized
- current date/time context conflicts with the manifest
- write scope is unclear or outside the manifest
- issue links are missing for tracked work
- selected track is not authorized
- token cap would be exceeded
- a source ledger is missing before synthesis
- a major contradiction has no resolution path
- live connector implementation is requested during a research-only phase
Output
End bootstrap with:
# OpenClaw Bootstrap Result
Date/time:
Manifest path:
Authorized track:
Write scopes:
Token cap:
Issue links:
Ready to launch:
Stop reason, if any:
Exact next command: