orchestrate-work

Coordinates repository delivery by delegating pinned task contracts to native agents.

Updated Jun 28, 2026
One-click install
npx skills add https://github.com/AO-HyS/aohys.com --skill orchestrate-work-ao-hys
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: orchestrate-work
Source: https://github.com/AO-HyS/aohys.com/tree/main/.agents/skills/orchestrate-work
Command: npx skills add https://github.com/AO-HyS/aohys.com --skill orchestrate-work-ao-hys

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Coordinating multi-agent work on a repository often fails because tasks are vaguely scoped, workers overlap on the same files, tests get modified without review, and completion is claimed without evidence. This Skill pins an explicit execution contract before any delegation so delivery finishes with verified, observable results. ## Core Features & Use Cases - Pinned execution contracts: Define owned files, delegation mode (exact instructions or outcome), expected behavior, stop conditions, and required checks before handing work to any agent. - Native agent delegation: Select workers based on actual runtime capabilities, run parallel packets only for disjoint ownership, and reuse the same worker for focused corrections. - Test-change policy enforcement: Block unauthorized test modifications through a shared runtime check against a reviewed exception policy. - Use Case: A lead engineer resumes a stalled feature branch, pins the contract for two disjoint worker packets, receives completion receipts with evidence, runs the named final checks, and reports the integrated result with remaining gaps. ## Quick Start Ask the agent to orchestrate the current ticket by pinning the execution contract, delegating disjoint work packets to available agents, and reporting verified completion evidence.

Frequently Asked Questions about orchestrate-work

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I coordinate multiple AI agents on the same repository?▼

Pin an execution contract for each worker specifying owned files, delegation mode, expected behavior, and stop conditions. Start parallel packets only when ownership is disjoint and dependency order is explicit, then integrate results as the parent orchestrator.

What is an execution contract in agent orchestration?▼

An execution contract defines the owned file surface, exact-instructions or outcome delegation mode, expected public behavior, authorization and stop conditions, and required checks. Both delegation modes share the same constraints; outcome mode lets the worker choose its implementation while running every mandatory check.

Can AI agents modify test files during orchestrated work?▼

No, the default test-change policy allows no test additions, modifications, or deletions. Changes require a reviewed exception identifying the path, SHA-256 or null for deletions, reason, and issue, verified by the findTestPolicyViolations runtime check.

When should I delegate work to a subagent versus doing it directly?▼

Delegate only when it shortens the critical path or supplies independent judgment. The parent retains decomposition, integration, conflict resolution, and acceptance; workers receive packets with root, revision, owned surface, checks, and evidence shape.

What evidence should an orchestrated run report at completion?▼

Report the candidate, actual model and tool identity, completion receipt, executed checks, links, and remaining gaps. Keep source, local runtime, PR, preview, production, and acceptance claims separate rather than merging them into one status.