yeaight7

Full autonomous execution from idea to working code

Run large codebase migrations in reviewable local batches with codemods, checkpoints, and verification. Use when a wide refactor is too risky to ship as one monolithic change.

Run independent work in parallel with explicit wave boundaries, lightweight coordination, and bounded verification. Use when latency matters and the work does not need a persistent completion loop.

Self-referential loop until task completion with configurable verification reviewer

Coordinate multiple subagents through dependency-aware waves, stage handoffs, monitoring, and verification loops. Use when the work is too large or too mixed for a single agent.

Turn vague implementation requests into testable requirements before coding.

Use when optimizing agent runtime loops, card packs, MCP session lifecycle, tool-call count, or multi-agent orchestration patterns.

Use when diagnosing agent session history, interrupted tool loops, missing tool results, timing bottlenecks, or subagent trace correlation.

Use when validating rendered web pages, local dev servers, browser automation, screenshots, forms, auth sessions, or UI evidence with strict browser safety boundaries.

Use when designing or reviewing filesystem MCP access, path boundaries, allowed roots, method allowlists, and safe local file operations.

Use when querying, ingesting, or maintaining a local RAG MCP corpus for semantic document retrieval with privacy controls.

Use when connecting to a managed codebase-context MCP/session service, checking stale maps, or safely using MCP-provided repository context.

Use when designing or using MCP-backed structured code search with search, AST query, symbol inventory, and bounded extraction workflows.

Audit agent configuration files for security vulnerabilities and misconfigurations. Covers settings.json, .mcp.json, .codex/config.toml, AGENTS.md, hooks, plugin manifests, and relay config. Classify findings as P0 or P1.

Design agent tool sets with stable names, narrow schemas, deterministic output shapes, and explicit error paths. No catch-all tools unless unavoidable.

Use when diagnosing agent session history, interrupted tool loops, missing tool results, timing bottlenecks, or subagent trace correlation.

Run an anti-slop cleanup workflow on AI-generated code — regression-tests-first, smell-by-smell, behavior preserved. Supports explicit review-only mode.

Use when a Claude Code agent needs to delegate a question, subtask, or review to a local Codex instance.

Use when a Codex or generic local agent needs a focused second opinion from Gemini CLI without MCP routing.

Full autonomous execution from idea to working code

You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.

Analyze BigQuery usage, identify cost hotspots, repeated failures, and practical optimization opportunities.

Use when validating rendered web pages, local dev servers, browser automation, screenshots, forms, auth sessions, or UI evidence with strict browser safety boundaries.

Use when reproducing, isolating, and fixing a bug with the smallest safe change.

Use when creating or reviewing red-team eval plugins, attack templates, grader rubrics, safety fixtures, or model-risk test metadata.

Fix broken builds, type errors, or test failures with the smallest safe diff — no opportunistic refactors, no cleanup, no unrelated changes.

Process-first advisor routing with artifact capture

Use when a Codex or generic local agent needs a focused second opinion from Claude CLI without MCP routing.

Use when connecting to a managed codebase-context MCP/session service, checking stale maps, or safely using MCP-provided repository context.

Use when optimizing agent runtime loops, card packs, MCP session lifecycle, tool-call count, or multi-agent orchestration patterns.

Relay questions and tasks to a persistent local Gemini ACP session with cross-turn context.

Use to audit and remove unused or redundant third-party dependencies from package manifests.

Use continuously during long tasks. Teaches how to read less, output less, and keep the LLM context window lean and fast.

Check existing repo capability, external libraries, MCP options, and maintenance risk before writing custom code. Decide adopt/wrap/build with explicit criteria.

Run an anti-slop cleanup workflow on AI-generated code — regression-tests-first, smell-by-smell, behavior preserved.

Use when designing or using MCP-backed structured code search with search, AST query, symbol inventory, and bounded extraction workflows.

Use when evaluating prompts, LLM outputs, red-team suites, or model behavior with local eval configs and safe provider/cost controls.

Deterministic 3-cycle loop for gathering codebase context before acting. Broad search → exact source and tests → target-specific docs and setup. Stop after enough context or report what is still missing.

Multi-model orchestration — route to two external advisors, then synthesize

Record why an architectural choice was made to prevent agents or humans from unintentionally reverting it.

Transforma commits, PRs ou diffs em notas de lançamento para usuários ou um changelog interno, com modos para detecção de mudanças que quebram a compatibilidade.

Run large codebase migrations in reviewable local batches with codemods, checkpoints, and verification. Use when a wide refactor is too risky to ship as one monolithic change.

Use when conversation context is bloated, constraints are being forgotten, or a compact handoff is needed before continuing work.

Use when designing, running, debugging, or hardening deterministic eval suites for agent skills, prompts, tool workflows, or MCP-backed cases.

Design agent tool sets with stable names, narrow schemas, deterministic output shapes, and explicit error paths. No catch-all tools unless unavoidable.

Vendor-neutral routing guide for choosing the right model tier by task type. Mechanical work uses a smaller/faster model; implementation uses a standard model; architecture, security, and release audit use the most capable model.

Deterministic 3-cycle loop for gathering codebase context before acting. Broad search → exact source and tests → target-specific docs and setup. Stop after enough context or report what is still missing.

Use when querying, ingesting, or maintaining a local RAG MCP corpus for semantic document retrieval with privacy controls.