oil-oil/codex-team-mode / skills/team-mode
team-mode
VerifiedUse when a task may benefit from a bounded subagent for implementation, large-codebase discovery at task start, independent review, requested or pre-commit code simplification, or expert work on complex decisions, modeling, automation, or repeated failures. The main agent chooses whether to delegate and accepts the result. Do not use for simple questions or short work with no meaningful review or delegation.
| File | Kind | Size |
|---|---|---|
| README.md | markdown | 1.8 KB |
| SKILL.md | skill md | 5.2 KB |
| agents/openai.yaml | config | 410 B |
| references/custom-agents.md | markdown | 4.1 KB |
| references/evaluation.md | markdown | 2.3 KB |
| references/explore.md | markdown | 654 B |
| references/interactive-testing.md | markdown | 1.4 KB |
| references/simplify.md | markdown | 2.9 KB |
| scripts/current_model.pyexecutable | script | 3.4 KB |
| scripts/usage_by_model.pyexecutable | script | 36.0 KB |
references/evaluation.md
View on GitHub# Evaluate Team Mode
Read this only when the user asks to evaluate routing, model choice, cost, or practical value.
## Collect Evidence
1. Record the task goal, baseline, acceptance checks, and installed role-to-model mapping.
2. Check each child's actual role, model, and effort in the runtime trace. TOML files alone do not prove what ran.
3. Judge the returned artifact and the parent's inspection or rework, not the child's self-assessment.
4. Compare comparable work only. Do not create duplicate work just to fill a benchmark.
For local usage, run `python3 scripts/usage_by_model.py --task-id current --by-agent --by-session --json`. Separate uncached input, cached input, output, reasoning output, and estimated Standard credits. Local traces can omit sessions, and credit estimates are not invoices.
Record whether the fresh brief supplied the goal, source paths, scope, and checks the child needed. Missing evidence is a briefing issue before it is a model issue. Distinguish a completed lifecycle event from an actual final report and from main-agent acceptance. Inspect shared artifacts before retrying a failed or interrupted child.
## Interpret The Roles
- `Explorer` is useful only when the main agent explicitly needs a large-codebase file map. Measure whether it found the right primary files and reduced parent search.
- `Executor` is useful when scope and checks are clear. Measure accepted output, rework, and time saved through independent ownership.
- `Reviewer` should find material issues or confirm a meaningful risk was checked. A report with no findings is not automatically wasted work. Check a saved Markdown report when one was requested or useful.
- `ExpertAdvisor` should contribute an independently grounded plan or complete the assigned modeling or automation. Check the actual model and result before calling it stronger-model work.
Compare coordination cost, child usage, wall-clock effect, missed context, and final result. A fixed role's TOML model or effort takes precedence over spawn values; do not count a passed override request as an A/B trial without trace evidence.
Change routing or briefing before upgrading every role. Change a lasting profile only when repeated task-scoped evidence supports it, and update the template, installed file, documentation, and prior conclusions together.
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: 472k | Updated | Claude CodeCodex | Verified | 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 |