Top-level orchestration skill that MUST be invoked at the start of every agent turn. Identifies required skills, coordinates sub-agents, and ensures comprehensive task completion with verification. Use when this capability is needed.
Top-level orchestration skill that MUST be invoked at the start of every agent turn. Identifies required skills, coordinates sub-agents, and ensures comprehensive task completion with verification. Use when this capability is needed.
metadata
{"author":"garyocean428"}
Master Orchestration Skill
MANDATORY: Invoke This Skill First
Every agent turn MUST begin by applying this skill. This ensures:
Correct skills are identified and applied
Sub-agents are coordinated effectively
No task is left incomplete or unverified
When to Use
Always. This is the entry point for all work.
Before any planning, implementation, or review task.
When receiving any user prompt or continuation.
Orchestration Workflow
1. Task Analysis
Upon receiving a task or prompt:
Parse the request - Identify explicit and implicit requirements
Categorize the work type:
Planning / Architecture
Implementation / Coding
Review / QA / Verification
Documentation / Roadmap
Research / Investigation
Identify scope - Single file, module, system-wide, or cross-cutting
Genesis Rollout Doctrine (Authoritative)
If the work touches kernel lifecycle, spawning, rollback/start flows, or governance, treat docs/11-Genesis-kernel-upgrade/* as the target end-state:
FOR each task in prioritized_tasks:
1. Announce: "Applying skill: {skill_name} for {purpose}"
2. Execute skill instructions completely
3. Record outputs and issues discovered
4. If issues found → add to task backlog
FOR implementation tasks:
1. Plan → Red-team plan → Research → Refine (2 iterations)
2. Implement → Red-team implementation → Fix → Verify (2 iterations)
3. QA → Prove completion → Update roadmap
5. Issue Tracking Protocol
All issues discovered during any phase become tasks:
Issues related to current work → Address immediately
Issues unrelated but important → Add to roadmap backlog
No issue is ignored or forgotten
Format for logging issues:
- ID: ISSUE-{timestamp}
- Severity: Critical / High / Medium / Low
- Area: {component/file}
- Description: {what's wrong}
- Proposed Fix: {how to address}
- Status: Open / In Progress / Resolved / Deferred
6. Completion Verification
Before finishing ANY turn, you MUST:
Prove all claimed implementations work (run tests, show output)
Verify all acceptance criteria are met
Confirm no regressions in existing functionality
Update the master roadmap with progress and new issues (docs/00-roadmap/20260112-master-roadmap-1.00W.md)
Push changes to git (if not already done)
List any deferred items with rationale
The completion statement must include:
Commit hashes for changes made
Test results summary
Explicit mapping: task → verification evidence
Any gaps or deferred work (with tracking references)
Integration with Existing Skills
This skill coordinates with all 22+ existing skills:
Always Active (invoke automatically)
qig-purity-validation - Every code change
dependency-management - Every dependency touch
e8-architecture-validation - Every architecture change
Invoke Based on Task
multi-agent-red-team-planning - For planning phases
multi-agent-red-team-implementation - For implementation phases
