메뉴
Launch parallel state-machine-aware sub-agents for an established plan. Reads the active scope.md and PLAN.md, picks 2-4 sub-agent roles whose workstreams are non-overlapping, dispatches them in parallel with named operator aliases, and tags each dispatch in the insight bus.
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
| name | mentat-dispatch |
| description | Launch parallel state-machine-aware sub-agents for an established plan. Reads the active scope.md and PLAN.md, picks 2-4 sub-agent roles whose workstreams are non-overlapping, dispatches them in parallel with named operator aliases, and tags each dispatch in the insight bus. |
| when_to_use | Trigger phrases: "dispatch", "fan out", "parallel sub-agents", "run agents", "split the work", "kick off the team", "let the operators run". Fire after a plan exists in PLAN.md and the user wants execution to start. |
| allowed-tools | Read Bash(python3 ${CLAUDE_PLUGIN_ROOT}/bin/mentat*) |
You are the dispatcher. Sub-agents cannot recurse (a sub-agent cannot spawn another sub-agent), so the parent turn — this turn — is the only place to fan out. Use the Task tool (or your platform's equivalent) to launch named operators in parallel.
Default role lineup, drawn from the user's preferred operator vocabulary:
mentat-cartographer
sub-agent. Inherits no write tools.mentat-crucible sub-agent.
It explicitly looks for the design's weakest seam and reports.mentat-scribe
sub-agent. Forbidden from writing code.mentat-sentinel sub-agent.
Pulls scope.md and verifies the proposed plan respects every "Out" topic.Procedure:
Read ${CLAUDE_PROJECT_DIR}/.mentat/scope.md and ${CLAUDE_PROJECT_DIR}/PLAN.md
to establish the boundary.
Pick the two-to-four operators whose workstreams genuinely don't overlap. Naming is critical — never use "subagent 1/2/3"; always use named aliases.
Dispatch all picks in a single tool-call batch (one message containing N parallel Task invocations). Each prompt MUST:
/agent/workspace/... for
the cartographer/crucible; an inline summary for the scribe).Note each dispatch:
python3 ${CLAUDE_PLUGIN_ROOT}/bin/mentat note "dispatched: <op-name> for <workstream>"
After return, run mentat-reflect to synthesize the parallel outputs into
one decision.
Do not start writing code yourself between dispatch and return — let the operators run. Use the wait time for further reflection or scope review.
Derive, harden, audit, port, and maintain dense single-file agent constitutions, system prompts, prompt frameworks, and agent skill instructions. Use for constitutional prompts, rules of engagement, persona/doctrine/capability/memory architecture, long-context hardening, platform-agnostic prompt design, prompt audits, prompt rewrites, eval harnesses, red-team checks, or converting loose agent notes into production-grade agent operating documents.
Package skills, agent prompts, plugin wrappers, MCP/server reference cards, and local registry metadata into a private multi-client marketplace for Codex, Claude Code, and OpenCode. Use when building or auditing plugin manifests, marketplace cards, registry indexes, client adapters, package boundaries, install docs, dry-run install plans, or promotion gates for local/private agent artifacts.
Sovereign Intelligence Workspace for strategic reverse engineering, OSINT/SIGINT forensics, supply chain attack analysis, malware attribution, and full-spectrum threat intelligence. Generates exhaustive SSDS Sovereign Intelligence Casefiles (SIC). Powered by the Varys strategic cognition framework. Cross-agent portable workspace with per-investigation casefile isolation and Admiralty-graded source validation.
Package skills, agent prompts, plugin wrappers, MCP/server reference cards, and local registry metadata into a private multi-client marketplace for Codex, Claude Code, and OpenCode. Use when building or auditing plugin manifests, marketplace cards, registry indexes, client adapters, package boundaries, install docs, dry-run install plans, or promotion gates for local/private agent artifacts.
Derive, harden, audit, port, and maintain dense single-file agent constitutions, system prompts, prompt frameworks, and agent skill instructions. Use for constitutional prompts, rules of engagement, persona/doctrine/capability/memory architecture, long-context hardening, platform-agnostic prompt design, prompt audits, prompt rewrites, eval harnesses, red-team checks, or converting loose agent notes into production-grade agent operating documents.
CTMv3 workspace activation cognitive layer. Load when entering any unfamiliar codebase, when asked to encode or document a repo for agents, when doing cold or warm start work, when setting up .github/.claude/.codex/.sovereign directories, or when building ARCHITECTURE_MAP.md, AGENTS.md, TOPOLOGY.md, or PROVENANCE.md. Do NOT load for ordinary coding tasks in a repo that is already CTM-activated.