Skip to content

foxrun-ai/foxrun / skills/orchestrate

orchestrate

Flagged

The one menu for handing work between Claude, Codex, and Hermes. Say where you are, what you want (quick hit, second opinion, review, fix, grill, live conversation, steer, parallel workers), and an optional session id; get one route back. Use whenever an agent needs another runtime — before Codex Desktop native tools, herdr, claude -p, codex exec, or hermes -z. Replaces fleet, foxy, dex, claude-delegate, and the bundled Hermes codex / claude-code skills.

tests/acceptance.md

View on GitHub
# Codex Desktop native delegation acceptance

These procedures create desktop state. T159 does not run them. Each is **TO RUN BY THE PERSON'S LEAD**.
The evidence note keeps this experiment out of the tracker, so each procedure expects zero tracker rows.

## Conversation during work

- Steps: Start C with a visible worker and valid return thread; continue two lead messages; steer the worker once; wait for its callback.
- Expected tracker rows: zero.
- Expected messages: two normal lead replies, one worker steering acknowledgement, then one completion callback with outcome, evidence, or blocker.
- Stop: stop after the first valid callback; stop earlier on wrong thread, expanded authority, or conflicting instructions.
- Result: NOT RUN — to run by Person's lead; no desktop thread was created.

## Callback delivery

- Steps: Start C with both thread IDs; require one callback; let the lead turn end; verify the callback appears in the named lead thread.
- Expected tracker rows: zero.
- Expected messages: exactly one visible completion callback with outcome, evidence, and any blocker.
- Stop: reconcile delivery before any retry; stop on the first valid callback or confirmed delivery failure.
- Result: NOT RUN — to run by Person's lead; no desktop thread was created.

## Worker failure

- Steps: Start C with a controlled failing command; require the worker to report failure without fixing outside its allowed rounds.
- Expected tracker rows: zero.
- Expected messages: exactly one callback naming failure, evidence, attempted allowed fixes, and the blocker.
- Stop: stop at the stated fix-round limit; do not start a replacement worker without direction.
- Result: NOT RUN — to run by Person's lead; no desktop thread was created.

## Duplicate reply

- Steps: Start C; after one valid callback, simulate an uncertain acknowledgement; inspect the lead thread before any resend.
- Expected tracker rows: zero.
- Expected messages: one completion callback only; no second callback after reconciliation finds the first.
- Stop: never retry blindly; stop if delivery cannot be reconciled and report uncertainty without sending again.
- Result: NOT RUN — to run by Person's lead; no desktop thread was created.

## Scheduled-check cleanup

- Steps: Create one D check with timing, evidence, and a stop rule; satisfy the condition; verify its final message and removal or disablement.
- Expected tracker rows: zero; one temporary native schedule during the test and zero active schedules after cleanup.
- Expected messages: one condition report, then one cleanup confirmation; no later empty check.
- Stop: stop and clean up when the condition is met, the run limit is reached, or evidence access fails.
- Result: NOT RUN — to run by Person's lead; no schedule was created.
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: 471kUpdated Claude CodeCodexVerifiedInstall
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: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
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 availableStars: 322kUpdated Claude CodeCodexVerifiedInstall