Squad MCP
未认领MCP server + Claude Code plugin: multi-agent advisory squad with PRD-decomposed tasks, persistent learnings, weighted rubric scorecard, and PR posting.
01 / 选择客户端
03 / 添加配置
claude mcp add --transport stdio 'squad-mcp' -- 'npx' '-y' '@gempack/squad-mcp'04 / 在客户端中确认
打开客户端的 MCP 设置，确认服务器已连接并列出工具。复制配置不代表连接成功。
更多AI & ML服务器
浏览完整目录安全概况
已认领和已认证的服务器每周接受一次静态扫描，显示代码能触及的范围（外部服务、环境变量、shell 命令、代理配置目录）以及带有已知公告的依赖。认领此列表即可获得。 安全概况的工作原理
26 个工具中显示 26 个
文档中列出的工具 (26)
内容来自项目文档。服务器握手不会验证每个工具的说明或行为。
apply_consolidation_rules
Aggregate advisory reports → final verdict (APPROVED / CHANGESREQUIRED / REJECTED). Returns weighted rubric scorecard when reports carry per-dimension scores.
classify_work_type
Heuristic WorkType from prompt + paths (Feature / Bug Fix / Refactor / Performance / Security / Business Rule) with Low/Medium/High confidence.
compose_advisory_bundle
One-call full bundle: composesquadworkflow + slicefilesforagent per selected agent + validateplantext.
compose_prd_parse
Build a prompt + JSON schema for the host LLM to decompose a PRD into atomic tasks. Pure-MCP: server does NO LLM calls. Caller (skill) feeds the prompt to its model, then calls recordtasks after user confirmation.
compose_squad_workflow
One-call pipeline: detectchangedfiles → classifyworktype → scorerisk → selectsquad.
detect_changed_files
Hardened git diff --name-status --no-renames for a workspace. Allowlisted refs, 10s timeout, 1MB stdout cap.
expand_task
Append subtasks to an existing task. Mechanical only — caller (skill or LLM) supplies the subtask inputs.
get_agent_definition
Return the full markdown system prompt for an agent (local override → embedded default).
init_local_config
Copy embedded defaults to the local override directory so they can be edited.
list_agents
List configured agents with role, ownership, naming conventions.
list_runs
Read-only journal read. Folds the two-row pair by id, filters (since / limit / agent / verdict / mode / invocation / worktype), and returns either the folded list or an aggregate bundle (outcomes + health + trend) when aggregate: true.
list_tasks
Read tasks from .squad/tasks.json. Filters: status, agent (matches agenthints), changedfiles (glob match against task scope).
next_task
Pick the next ready task: candidate status (default pending), all dependencies done, optional agent / changedfiles filter. Tiebreak priority then id. Returns null + reason when none ready.
prune_learnings
Lifecycle maintenance (v0.11.0+): mark entries older than maxagedays as archived and entries with ≥ minrecurrence accepts on the same canonicalised finding as promoted. Atomic rewrite under file lock. Never auto-runs.
read_learnings
Load past accept/reject decisions from .squad/learnings.jsonl. Filters by agent / decision / changed-file scope. Returns entries plus a markdown block ready to inject into agent or consolidator prompts.
read_squad_config
Read and resolve .squad.yaml (or .squad.yml) at workspaceroot. Returns effective weights, threshold, minscore, skippaths, disableagents.
record_learning
Append one accept/reject decision to .squad/learnings.jsonl. Side-effecting; the skill (or CLI) is responsible for per-finding user authorisation.
record_run
Append one RunRecord to .squad/runs.jsonl. Single-writer contract: only the squad skill calls this (Phase 1 inflight + Phase 10 terminal). Validates against schemaversion 1, enforces 4 KB per-record cap, file mode 0o600.
record_tasks
Bulk-create tasks. Allocates ids sequentially, validates dependencies resolve (forward refs in batch ok), rejects duplicates and self-deps. Atomic write.
score_risk
Compute Low/Medium/High from boolean signals (auth, money, migration, filescount, newmodule, apichange).
score_rubric
Pure rubric calculator. Takes per-agent scores (0-100) + optional weight overrides, returns weighted score, per-dimension breakdown, and pre-formatted ASCII scorecard.
select_squad
Select advisory agents for a work type. Combines matrix + path hints + content sniff. Returns evidence per file.
slice_files_for_agent
Filter a file list to those owned by a single agent. Used to build sliced advisory prompts.
slice_files_for_task
Filter a file list to those matching a task's scope glob. Same glob primitive as skippaths and learnings scope.
update_task_status
Flip a task or subtask status: pending / in-progress / review / done / blocked / cancelled.
validate_plan_text
Advisory check for inviolable-rule violations in a plan (commit/push fences, emojis in code blocks, non-English identifiers, impl-before-approval).