| name | openskills-dev-orchestrator |
| description | Route OpenSkills development tasks to the right project skill or subagent, including sequencing rules for debugging, feature work, regression checks, and release readiness. |
OpenSkills Dev Orchestrator
Use this skill to decide which OpenSkills project skill or subagent to invoke first, and in what order, for the fastest safe outcome.
Routing Rules
1) Runtime failures or sandbox regressions
- Start with
openskills-runtime-debug.
- If issue spans multiple execution paths or platforms, use
runtime-sandbox-auditor subagent.
- Re-run runtime tests before concluding.
2) Plugin/build feature changes
- Start with
openskills-plugin-separation.
- If dependency topology is unclear, use
wasm-plugin-build-specialist subagent.
- Validate runtime default-feature behavior and binding impact.
3) Binding breakages (TS/Python)
- Start with
openskills-bindings-maintainer.
- If both bindings are affected or unclear, use
bindings-compatibility-agent subagent.
- Return a compatibility matrix and migration notes if needed.
4) Skill package authoring or updates
- Start with
openskills-skill-authoring.
- Use
skill-spec-conformance-agent subagent when multiple skills or ambiguous triggers are involved.
- Validate discovery and activation behavior.
5) End-to-end confidence checks
- Start with
openskills-e2e-test-runbook.
- Use
examples-e2e-agent subagent for scenario-heavy or flaky behavior investigations.
- Capture scenario-by-scenario pass/fail and tool-call evidence.
6) Release preparation
- Start with
openskills-release-ops.
- Use
release-gatekeeper-agent subagent for final GO/NO-GO judgment.
- Output blockers, risks, and rollback notes explicitly.
Escalation Heuristics
- Use a skill first when the workflow is deterministic and repeatable.
- Escalate to a subagent when the task needs deep cross-file reasoning, ambiguity handling, or parallel analysis.
- If a task is unresolved after one full skill pass, escalate to the corresponding subagent.
Output Contract
Always return:
- Selected path (skill and/or subagent)
- Why this path was chosen
- Ordered execution plan
- Success criteria
