메뉴
Delegate code generation tasks to Codex CLI for optimal performance. Use when implementing features, generating tests, refactoring code, or making bulk code changes. Triggers on "delegate to codex", "route to codex", "use codex", "code generation".
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Operating manual for documentation work in nexus-agents. Use when updating docs, adding new docs, changing doc pipeline, or troubleshooting doc issues. Triggers on "update docs", "add documentation", "doc pipeline".
Periodically exercise ALL nexus-agents feature families end-to-end against a real task — research → synthesize → vote → plan → dev-pipeline → graph workflow → memory → audit — capturing ACTUAL outputs to validate that fixes, docs, and claims hold in real usage (the things unit tests miss: live voter panels, adapter routing, pipeline stage wiring, audit chains). Use after a release, weekly, when several fixes have landed since the last run, or on demand. Triggers on "e2e validation", "end-to-end validation", "dogfood the loops", "validate end to end", "real usage test", "periodic validation run", "full-loop validation".
Run the CI gates locally before pushing, so failures surface in seconds instead of one CI cycle at a time. Enumerates the checks in .github/workflows/, runs the locally-runnable subset in CI's order, and pins the local-vs-CI delta as a memory the first time you work in a repo. Use before `git push` / opening a PR, or when CI just failed on something your local gate missed. Triggers on "pre-push", "before push", "ci parity", "will ci pass", "local gate", "prepush", "pre-push check".
Review code changes following project standards and security guidelines. Checks for lint compliance, type safety, test coverage, and security issues. Use when reviewing PRs, auditing code, checking implementation quality, or running a blind pre-merge review pass before merging your own change. Triggers on "review code", "code review", "check this", "audit", "PR review", "pre-merge review", "blind reviewer", "before merge".
Execute a release following project standards. Use when publishing a new version, creating release tags, or deploying. Triggers on "release", "publish", "version bump", "create release".
Improve an existing technical doc in-place via a phased Audit → Research → Rewrite → Validate workflow. Calls docs-review for audit and validate; never re-derives scoring. Applies answer-first H2s, citation capsules, and info-gain markers per .rules/docs-rubric.md. Use when the user says "rewrite docs", "improve this doc", "optimize doc", "docs rewrite".
| name | codex-delegator |
| description | Delegate code generation tasks to Codex CLI for optimal performance. Use when implementing features, generating tests, refactoring code, or making bulk code changes. Triggers on "delegate to codex", "route to codex", "use codex", "code generation". |
| allowed-tools | Bash, Read, Glob, Grep, Task |
Full documentation:
| Metric | Value | Notes |
|---|---|---|
| Model version | gpt-5.2-codex | Latest research preview |
| Latency (simple) | 3-4 seconds | Single function generation |
| Latency (complex) | 5-6 seconds | Interface + class boilerplate |
| Token overhead | ~900-1050 per session | Fixed session initialization cost |
| Default sandbox | read-only | Safe for code generation |
| Code quality | Production-ready | Proper generics, types, error handling |
Codex excels at:
| Task Type | Examples |
|---|---|
| Code Generation | Implement functions, create classes, add features |
| Test Generation | Write unit tests, integration tests, test fixtures |
| Refactoring | Extract methods, rename variables, restructure |
| Code Completion | Complete partial implementations, fill in stubs |
| Bulk Code Changes | Apply patterns across multiple files |
Do NOT use for: Architecture decisions, security review, documentation-heavy tasks.
Use the delegate_to_model MCP tool for intelligent routing:
delegate_to_model(task: "Implement a binary search function in TypeScript")
The CompositeRouter automatically routes to Codex when:
For explicit Codex invocation, use codex exec for non-interactive mode:
# Basic code generation (non-interactive)
codex exec "Implement a binary search function in TypeScript"
# Via nexus-agents CLI
nexus-agents orchestrate "Implement sorting algorithm" --cli=codex
# Explicit model selection (default is gpt-5.2-codex)
codex exec -m gpt-5.2-codex "Implement a debounce function"
# Full autonomous mode (dangerous - writes files)
codex exec --full-auto "Add error handling to all functions in utils.ts"
Important: Use codex exec NOT codex -p. The -p flag is for interactive sessions.
Check routing decisions before execution:
nexus-agents routing-audit "Implement authentication middleware" --format=json
The router analyzes tasks using these signals:
| Signal | Boost Codex When |
|---|---|
codeGeneration | Keywords: implement, write, create |
reasoningSimple | No "design", "architect", "analyze" |
budgetSensitive | Keywords: quick, simple, straightforward |
technicalContext | References to APIs, frameworks, libs |
delegate_to_model(task: "Write Jest tests for the UserService class covering:
- createUser with valid input
- createUser with invalid email
- getUserById with existing user
- getUserById with non-existent user")
delegate_to_model(task: "Implement a rate limiter middleware for Express.js that:
- Uses sliding window algorithm
- Supports configurable limits per route
- Returns 429 status when limit exceeded")
delegate_to_model(task: "Refactor the payment processing module to:
- Extract validation logic into separate functions
- Replace callbacks with async/await
- Add TypeScript types for all parameters")
delegate_to_model(task: "Add JSDoc comments to all exported functions in src/utils/")
Task Input
|
v
TaskAnalyzer (profile task)
|
v
BudgetRouter (filter by constraints)
|
v
TopsisRouter (rank: quality 50%, cost 30%, latency 20%)
|
v
LinUCBBandit (contextual learning)
|
v
Codex selected if: high codeGeneration + low reasoningComplexity
| Capability | Score | Notes |
|---|---|---|
| Code Generation | 0.95 | Primary strength |
| Context Window | 400K | Large file support |
| Cost Efficiency | High | Lower than Claude for code tasks |
| Latency | 3-6s | Fast for focused code tasks |
tsc --noEmit on TypeScript before integrationIf Codex is unavailable:
# nexus-agents.yaml
routing:
enableTopsisRanking: true
enableLinUCBSelection: true
topsis:
qualityWeight: 0.5
costWeight: 0.3
latencyWeight: 0.2
# Check Codex availability
nexus-agents doctor
# View routing decision for task
nexus-agents routing-audit "task description" --verbose
# Show bandit learning stats
nexus-agents routing-audit "task" --bandit-stats
| Excuse | Counter |
|---|---|
| "Codex is fastest, always use it" | Codex is fastest for code generation. For research/docs, Gemini wins on context; for design, Claude wins. Match the CLI to the task type. |
| "I'll route everything through Codex" | Performance floors fire when one CLI is overweighted. Spread load per the routing system (CompositeRouter). |
| "Skip the model probe" | Required after init (#1402). Without it, you may try to use a model the CLI doesn't expose. |
| "Codex failed, retry with same params" | Transient retry already runs (#1531). If it failed twice, the failure is real — fall back to a different CLI per the routing chain. |
CompositeRouter.route() — no direct adapter instantiation