richfrem/agent-plugins-skills / plugins/agent-orchestration/.apm/skills/orchestrating-agent-work
orchestrating-agent-work
VerifiedInteractive orchestrator for complex engineering work. Analyzes the task, conducts a diagnostic interview to select the appropriate execution strategy (Directed Acyclic Graph versus Loop, and which Loop type), plans the artifact, and supervises execution and verification. Use for "how should I orchestrate this", "plan this workflow", "should this be a DAG or a loop", "run these jobs in parallel", "delegate this to a worker agent", or "run/resume this graph manifest". Not for single trivial edits or skill self-improvement.
references/dispatch-preflight.md
View on GitHub# Dispatch preflight and completion discipline
Observed in a four-worker batch (Sanctuary MISSION-052). Each rule says where it is enforced; the guards live in `cli-agent/scripts/dispatch_guard.py` and run inside `run_agent.py` when its flags are passed.
| Failure | Rule | Enforced by |
|---|---|---|
| Worker returned a partial slice and it was treated as done | Done = the assigned OUTCOME (for experiments: a verdict or a recorded "cannot decide"), not worker return. Redispatch defects. | acceptance-criteria.md; judgment |
| Independent review waited for all workers | Review each ready delivery while other independent work keeps running. | judgment (schedule it explicitly) |
| Cloud arm idled behind a congested local resource | Start independent arms at once; schedule the shared resource explicitly and record its effect on validity. | judgment |
| Duplicate execution detected only afterward | Claim the unit before launch; a failed claim means do not launch. | `dispatch_guard.py claim` |
| Worker lacked a required executable | Check in the worker's own environment (its PATH and sandbox), before dispatch. | `dispatch_guard.py exec-check --path <worker PATH>` run by the worker or its wrapper |
| Hundreds of calls ran with hooks, memory and CLAUDE.md loaded | Verify the call context from the CLI's init event before any large run; refuse a contaminated invocation; record what is unobservable. Do not pool contaminated and clean outputs. | `dispatch_guard.py context-check` on a stream-json probe |
| Progress counted log rows | Report unique planned/completed/malformed/duplicate/missing units. Malformed answers stay in the denominator. | `dispatch_guard.py reconcile` |
| Permission asked again inside a standing authorization | Quote the authorization and act within its scope; ask only for new scope or a real blocker. | judgment |
| Telemetry showed zero errors or corrections despite observed ones | Report a telemetry gap; never treat missing as zero. | judgment; Sanctuary `eval_gate.py invest` |
A held lock or shared resource is announced to the affected worker before it is held. Give each execution unit one owner.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Unverified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |