cross-model-delegation

Plugin cross-model-delegation from marketplace cross-model-agent-delegation in alectivism/cross-model-agent-delegation. Commands read from the repo's manifest — today.

Skills (3)

• claude-accounts

  Run a second Claude Code login (for example work and personal) side by side with one shared configuration, move a conversation from one login to the other, or run a task on the second login from the first. Use for "switch accounts", "use my other account", "claude-personal", the ACCOUNTS line, or se

• cross-model-delegation

  Use when delegating any task to another model family. Codex CLI (codex exec, GPT-6, "ask Codex", offload to GPT) for code, review and reasoning, or Gemini 4 Argon through the Antigravity CLI (agy, "ask Gemini") for writing. Covers cross-model review, second opinions, choosing a Codex model/effort ti

• session-control

  Run slash commands in this Claude Code session yourself through its cmux pane. Use to compact at a good moment (a CONTEXT line appears, or the user says "compact"), to move this conversation to another Claude login (an ACCOUNTS line shows switch=<name>, or "switch accounts"), or to turn an MCP serve

Subagents (10)

• agent-name

  One-line description: what it does AND when Claude should delegate to it (Claude routes on this).

• adversarial-verifier

  Independent read-only verifier for another agent's (or model's) output — a diff, a claim, extracted data. Tries to REFUTE, never to bless. Use as the second half of a generator/verifier split, ideally on a different model than the generator.

• bug-investigator

  Investigates bugs by tracing execution paths, reading logs, checking recent changes, and identifying root causes. Use when the cause isn't obvious from the error message alone. Diagnosis only; returns ranked fix options.

• content-writer

  Long-form content that needs research and writing together (posts, guides, case studies). Keeps research and drafting in one context to prevent the telephone-game information loss of splitting them.

• deep-research

  Deep multi-source research generating significant intermediate context. Use for topics needing extensive exploration (5+ searches, multiple angles) before a clean summary is possible. Can spawn its own per-source readers.

• dependency-auditor

  Checks project dependencies for outdated versions, known vulnerabilities, and license issues. Use before releases or when adopting a project. Mechanical scan-and-report.

• doc-writer

  Generates or updates documentation from code: READMEs, API docs, docstrings, setup guides. Use after code changes leave docs stale, or when a project lacks docs.

• pr-preparer

  Generates PR title, description, and test plan from the current branch's diff against the base branch. Use before creating any pull request. Mechanical diff summarization; read-only.

• researcher

  Searches the web and local files and returns structured, dated, source-attributed findings. Use for any research task needing 3+ searches or multiple sources; the intermediate search noise stays out of the main context.

• test-writer

  Writes tests for new or changed code, following the project's existing test patterns, then runs them to prove they pass. Use after implementing features or fixes that lack coverage.

Similar plugins