Plugin Claude Code
Plugin marketplace listing 1 plugin: mcp-agent-relay.
0 19d ago A
tokens not measured
Plugin Claude Code
A durable MCP job relay for allowlisted Codex and Claude CLI executor profiles.
0 19d ago A
tokens not measured
agentrelay
03MCP server Claude CodeCodexCursor +2
MCP server "agentrelay" as configured in brenoperucchi/mcp-agent-relay. Runs ${CLAUDE_PLUGIN_ROOT}/server.mjs with node. Needs 4 environment variables to run.
0 19d ago A
tokens not measured
Instructions file CodexOpenCode
AGENTS.md instructions for brenoperucchi/mcp-agent-relay, covering repository guidance and agent relay wait policy.
0 19d ago A 370 tokens
implement
05Command
Dispatch an implementation task to Codex via the agentrelay relay, isolated in its own git worktree. Runs on the background worker; nothing is written to the main working tree or merged automatically.
0 19d ago A 39 tokens
review
06Command
Adversarially review a file with Codex via the agentrelay relay (dispatch, then wait for the verdict).
0 19d ago A 24 tokens
Agent
Use for reasoning-heavy phases — architecture decisions, debugging complex/subtle issues, algorithm design, evaluating trade-offs between approaches. Think thoroughly, but return a concise conclusion the orchestrator can act on directly, not a transcript of the reasoning.
0 19d ago A 52 tokens
Agent
A second, independent reasoning lens — run in parallel with deep-reasoner (and codex, when available) on plan reviews and high-stakes decisions, never as a solo replacement for deep-reasoner. The value is a different model family scrutinizing the same problem, not a faster or cheaper deep-reasoner.
0 19d ago A 72 tokens
fast-worker
09Agent
Use for mechanical, well-specified work — boilerplate, repetitive edits across files, writing tests for already-decided behavior, formatting, straightforward refactors. Not for anything requiring judgment calls or architecture decisions; that goes to deep-reasoner instead.
0 19d ago A 53 tokens
codex-i
10Command
Envia uma tarefa de implementação pro Codex via MCP agent relay (dispatchwait), usando gpt-5.6-sol/high e worktree própria. Roda no worker em background; NADA é escrito no working tree principal nem mergeado automaticamente.
0 19d ago A 51 tokens
codex-r
11Command
Envia uma revisão de código pro Codex via MCP agent relay (dispatchwait), usando gpt-5.6-sol/high, e devolve o veredito. Roda no worker em background, sem websocket direto.
0 19d ago A 46 tokens
orchestrate
12Command
Inicia o workflow de orquestração multi-agente (deep-reasoner=alias Opus atual, fable-reasoner=Fable 5, fast-worker=Sonnet, codex via MCP agent relay) para uma necessidade ou plano — cobre revisão do plano em paralelo e/ou delegação da execução.
0 19d ago A 66 tokens
dual-r
13Skill Claude CodeCodex
Revisão read-only por duas lentes independentes, com roteamento host-aware — Sol xhigh e Opus max — e evidência explícita de modelo.
0 19d ago B 36 tokens
Agent
Independent read-only Opus max reviewer for the dual-r panel; challenge invariants and architecture.
0 19d ago A 26 tokens
tri-r
15Skill Claude CodeCodex
Revisão de alta confiança por três famílias de modelo independentes — Opus, Fable e GPT-5.6 Sol — com análise cega em paralelo, quadro de divergência, rodada adversarial cruzada e verificação factual. O roteamento se adapta ao host Claude Code ou Codex sem confundir nome de agente com identidade de modelo.
0 19d ago A 75 tokens