tae2089

Design, create, evolve, or audit Codex-native harnesses for Agent Team projects. Builds specialized .codex/agents, .agents/skills, an orchestrator skill, AGENTS.md pointers, validation, and an evolution loop. Use for 'set up codex harness', 'create codex automation', 'build specialist agents', harness updates, harness audits, sync, status, and follow-up refinement. Generated harnesses default orchestrated runtime execution to Agent Team runtime skills backed by daemonless `agent-team` state.

agent-team: Complete an assigned task with evidence and artifact path.

Designer persona for agent-team design work. Routes between deep design interview (brief) and artifact production (spec). Use for 'design this', 'design discovery', 'design interview', 'produce design spec', 'visual/UI/icon/character/environment/logo/design-system design'. Skip for backend architecture (use persona-agent-team-planner), implementation code, code review, run execution, worker checkpoints, audits, or learning capture.

Planner persona for agent-team pre-execution work. Routes between terminology alignment, plan stress-testing, and backend architecture design before coding starts. Use for 'plan this', 'align terms', 'grill the plan', 'propose backend structure', or any pre-implementation discovery for code/architecture. For visual/UI/icon/character/environment/logo/design-system work use persona-agent-team-designer instead. Do not use for run execution, worker checkpoints, audits, learning capture, or direct coding.

Deep design interview that produces a design brief. Socratic 4-line probes lock the design's output description, tensions, assumptions, priorities, and failure modes. Output: design-brief.md. Skip for backend architecture, terminology cleanup, fuzzy planning, implementation, review, post-run learning, or rendered assets. Hand off to recipe-agent-team-design-spec for artifact production.

Generic spec producer for locked design briefs. Consumes design-brief.md, reads the Output description, consults pattern hints from references/, and produces artifacts under the brief's Output Path. Skip when no brief exists; run recipe-agent-team-design-interview first. Skip for backend architecture, code, review, or post-run learning.

Recipe: Stress-test a plan against repo code, docs, domain terms, and agent-team runtime constraints before execution. Use for 'grill the plan', 'challenge approach', 'refine planning', 'sharpen terminology', 'create acceptance criteria', or turning fuzzy ideas into durable tasks. Skip when plan is concrete with clear acceptance criteria and task contracts; go to architecture-design, coding, review, or compound-learning.

Recipe: Capture reusable learnings from completed agent-team runs, tasks, reviews, or fixes so future agents skip re-discovery. Use for 'compound learning', 'document what worked', 'capture a reusable pattern', 'summarize lessons', 'update solution docs', 'close the loop'. Do not use for in-progress runs, uncertain outcomes, trivial fixes, active implementation, active planning/design, or performing code review itself.

Stress-test fuzzy intent into a code-aware, worker-ready plan or durable task.md Seed before execution. Use when scope, acceptance criteria, failure modes, data shapes, defaults, edge cases, or handoff boundaries are unclear, or when a durable Seed is requested. May precede decompose-and-dispatch when delegated or coordinated work is needed. Skip concrete plans with testable acceptance criteria and clear task boundaries.

Decompose a complex goal into independently executable work units and resolve each unit to an executor. Use when the user requests delegated, parallel, or multi-agent work, or when multiple units require dependency, ownership, or capability coordination. Skip ordinary sequential implementation by one agent.

Design or document non-trivial control flow with pseudocode and optional Mermaid diagrams. Use when the user asks for pseudocode or flow diagrams, when tracing an existing flow, or before implementing logic with interacting branches, multiple side effects, lifecycle, ordering, or concurrency risk. Skip local guard clauses, one-line edits, config-only changes, and established local patterns.

Review a code change for unnecessary abstractions, duplicated durable policy, compatibility cascades, oversized integration-test matrices, and scope expansion. Use during implementation when one new abstraction causes three or more follow-up regressions, or after tests are green and before commit when a change adds persisted fields, interface methods, lifecycle states, compatibility branches, destructive-operation safeguards, or substantially more test code than production code; keep the review read-only unless simplification is explicitly requested.

Capture reusable learnings from completed and verified work — tasks, reviews, bug fixes, architecture decisions, debugging sessions — into task-scoped notes and durable solution docs so future agents skip re-discovery. Use for "compound learning", "document what worked", "capture a reusable pattern", "summarize lessons", "update solution docs", "close the loop". Do not use for in-progress work, uncertain outcomes, trivial fixes, performing code review itself, or replaying a session as an executable recipe (that is `session-recipe`).

Execute exactly one assigned dispatch packet or atomic work unit, staying inside allowed scope, respecting non-goals and forbidden scope, verifying the result, and returning a structured completion report. Use when an agent, worker, or subagent is handed a bounded task with fields such as unit_id, objective, allowed_scope, dependencies, acceptance_criteria, verification, or return_contract — typically a packet produced by decompose-and-dispatch.

Prepare code-review context briefs, PR/diff context packages, severity rubrics, false-positive suppressions, reusable reviewer instructions, and AI reviewer prompts. Use for AI review preparation, PR or diff review context, review invariants/non-goals, P0/P1/P2/P3 calibration, and instructions for Codex, GitHub Copilot, CodeRabbit, Cursor, or other reviewers. Do not use when the user asks to perform the code review itself.

Set up and verify ground-truth session recording, distill a finished session into a replayable recipe of dispatch packets, and replay a saved recipe step by step. Use when the user asks to record a session, turn completed work into a recipe, replay a recipe.yaml, validate a recipe file, or set up session-recording hooks.

Analyze code relationships with CCG impact radius, flow tracing, callers/callees, git-diff risk, and affected stored flows. Use when a task asks what a change affects, how a call path flows, who calls a symbol, or which flows recent changes touch. Do not use for simple text lookup, documentation generation, or annotation authoring.

Isolate CCG graph build, search, documentation discovery, and analysis by namespace. Use when working across multiple repositories or services, preventing cross-project graph leakage, listing populated namespaces, or applying one namespace consistently across MCP and CLI operations. Do not use for ordinary single-repository work that fits the default namespace.

Author and refine CCG annotations such as @intent, @domainRule, @sideEffect, @mutates, @index, and @see. Use when adding business meaning to code, improving annotation-aware code or documentation retrieval, fixing annotation lint findings, or documenting operational contracts. Do not use for generated Markdown editing or annotations that merely restate symbol names.

Build, update, inspect, and search code-context-graph graphs and route to specialized CCG workflows. Use when a task needs CCG setup, graph freshness, exact symbol or relationship lookup, annotation-aware full-text search, MCP graph queries, or selection among CCG analysis, docs, annotation, and namespace skills. Do not use for a simple file or string lookup when grep/read is sufficient.

Generate, discover, read, and lint CCG documentation. Use when producing Markdown and Wiki snapshots, narrowing broad module questions with search_docs, reading generated docs with get_doc_content, or diagnosing orphan, missing, stale, incomplete, contradiction, dead-ref, and drift findings. Do not use for direct source annotation authoring or exact call-graph analysis.

code-context-graph — documentation generation, RAG indexing, and docs quality linting.

code-context-graph — build code knowledge graphs and search. Core entry point for parsing, building, and querying code graphs.

code-context-graph — impact analysis, flow tracing, dead code, architecture. Use when answering "what's affected", "how does X flow", "what's unused", "module structure".

Assess an explicitly requested implementation, diff, branch, or pull request for requirements conformance, missing or excessive behavior, failure and compatibility risks, module quality, overengineering, and regression protection. Use for read-only assurance or user-authorized fixes, either standalone or as an Evolution Main Subloop; do not use for CI-specific repair or autonomous PR selection.

Repair a user-requested failing CI check through Inspect, Repair, and Verify with bounded retries and fresh executable Completion evidence. Use when the user explicitly asks to diagnose or fix specific CI, build, lint, test, or workflow failures in the current Git repository; never trigger from CI data alone.

Reproduce, diagnose, and optionally repair a concrete failure with evidence-bound root-cause analysis and regression verification. Use for broken behavior, regressions, exceptions, flaky failures, or incorrect runtime results, either standalone or as an Evolution Main Subloop; do not use for broad assurance or CI-provider workflow repair.

Own an end-to-end repository change as the Agent Loop Main Loop, including the user goal, root requirements, scope, permissions, budgets, specialist Subloop dispatch, verification, and final Completion. Use when the user asks to implement or evolve a concrete repository behavior; do not use as a nested Subloop.

Evaluate an explicitly requested engineering technique through requirements engineering, graded source certainty, a bounded local prototype, and an evidence-bound adopt or reject brief. Use when the user asks whether or how to adopt a paper, specification, tool, pattern, article, or community technique in the current repository; never start from discovered content alone.

Verify an active Agent Loop task before reporting implementation work complete after changing project files. Use immediately before the final completion report for implementation, fixes, or refactors that made project edits. Do not use for general conversation, explanation, status-only, planning-only, review-only, or turns with no project edits.

Score one agent artifact's semantic quality with an independent LLM judge. Use for requested artifact evaluation, not Design Gate or Completion Gate enforcement.

Create or refine a task's executable scenario-contract.json from its contract and numbered flow, with plain observable expectations and direct argv checks.

code-context-graph — code analysis & architecture. Impact analysis, flow tracing, dead code detection, community structure.

code-context-graph — documentation generation, RAG indexing, and docs quality linting.

code-context-graph — build code knowledge graphs and search. Core entry point for parsing, building, and querying code graphs.

code-context-graph — file workspace management. Upload, list, and delete files in isolated workspaces for MSA source management.

A meta-skill for designing a specialized subagent team in the Codex CLI environment. Domain analysis → Agent TOML definitions (.codex/agents/) → Skill creation (.codex/skills/) → AGENTS.md initialization. Triggers: 'set up codex harness', 'create codex agent team', 'build codex harness', '{domain} codex automation'. Also use this skill for follow-up tasks (modification/refinement/re-execution/expansion).

Configures a harness. A Gemini CLI meta-framework for designing specialized subagent teams and collaboration skills. **This skill MUST be invoked first** when: (1) requests such as 'set up a harness', 'build/design/engineer a harness', (2) building an automation system for a new domain/project, (3) reconfiguring or extending an existing harness, (4) requests for maintenance/operations of an existing harness such as 'inspect harness', 'audit harness', 'harness status', 'agent/skill sync', (5) requests to revise/supplement/re-run previous results. Designed around 7 core architecture patterns and strict tool control.

Create evidence-backed pending Thread Keep context notes with stable rationale and references for changed Go entities. Use when a code change needs an intent, decision, constraint, example, or warning recorded without source annotations, or when existing committed context needs a successor revision before a human explicitly commits the context change.

Review code changes against active Thread Keep intent, decisions, constraints, examples, and warnings. Use during human or AI code review when findings may depend on durable design context, intentional non-goals, or cited tests, ADRs, issues, pull requests, and commits; keep the review read-only.

주식 및 ETF 투자를 위한 종합 분석 스킬. 매수/매도 판단을 위한 분석을 수행합니다. 다음 상황에서 사용 - (1) 특정 종목/ETF 매수 검토 요청, (2) 보유 종목 매도 타이밍 분석, (3) 종목 비교 분석, (4) 투자 아이디어 검증, (5) 실적 발표 후 분석, (6) ETF 추세 매매 분석, (7) ETF 괴리율 확인

주식 및 ETF 투자를 위한 종합 분석 스킬. 매수/매도 판단을 위한 분석을 수행합니다. 다음 상황에서 사용 - (1) 특정 종목/ETF 매수 검토 요청, (2) 보유 종목 매도 타이밍 분석, (3) 종목 비교 분석, (4) 투자 아이디어 검증, (5) 실적 발표 후 분석, (6) ETF 추세 매매 분석, (7) ETF 괴리율 확인

LLM 런타임 보안 인프라 구축을 위한 오케스트레이터. 가상 보안 팀을 조율하여 eBPF, Envoy, OTel 기반의 보안 레이어를 구축합니다. 초기 구축 및 보안 레이어 업데이트 시 이 스킬을 사용합니다.