⚠️ INTERNAL USE ONLY — 请使用CC原生的TeamCreate工具创建团队，不要调用此MCP工具。

NOTE: For normal workflow, use CC's TeamCreate tool instead — it auto-registers the team via hooks. This MCP tool only creates a DB record without CC integration.

Get detailed information and status of a specified team.

List all created teams.

Returns: Team list with basic info for each team

Get a team panoramic briefing — understand full team status in one call.

Returns team info, member status, recent events, recent meetings, pending tasks, and action suggestions.

Close (complete) a team — sets team status to completed and marks all busy agents as offline.

Use this when the team's mission is fully done. Members are not deleted, but their status is set to offline automatically.

Delete a team.

Get recommended team role configuration based on project type.

⚠️ INTERNAL USE ONLY — 请使用CC原生的Agent工具创建Agent，不要调用此MCP工具。

NOTE: For normal workflow, use CC's Agent tool with team_name parameter instead. CC Agent tool spawns a real subprocess AND auto-registers via hooks. This MCP tool only creates a DB record — no actual agent process is started.

Update an Agent's running status.

List all registered Agents in a team.

List all available Agent templates (from ~/.claude/agents/).

Returns a template list and a grouped-by-category view to help choose the right Agent role template.

Returns: templates: All template list grouped: Templates grouped by category total: Total template count

Recommend suitable Agent templates based on task type and keywords.

Query Agent activity records for a team.

Returns recent activity log entries sorted by timestamp descending, including action type, duration_ms, and result summary.

Create a team meeting and return a ready-to-use dispatch_plan for spawning participant Agents.

Supports two participant formats:

1. Legacy (strings): participants=["arch-lead", "backend-arch"] Returns dispatch_plan with empty launch_call + deprecation warning.

2. Structured (dicts): participants=[{"name": "arch-lead", "agent_template": "software-architect", "role": "负责评估架构方案", "context_files": ["docs/arch.md"], "expected_output": "三段式"}] Returns dispatch_plan with fully populated launch_call.params ready to paste into Agent tool.

Send a discussion message in a meeting.

Discussion rules:

• Round 1: Each participant presents their views

• Round 2+: Must read previous speakers' messages first, cite and respond to specific points

• Final round: Summarize consensus and disagreements

SECURITY: Set caller_agent_id to the actual agent making this call. If it differs from agent_id, the message is flagged as impersonation in the audit log. Leader sending on behalf of others should set caller_agent_id='team-lead'.

Read all discussion messages in a meeting.

Conclude a meeting, marking it as completed.

By default checks that all expected participants have spoken before concluding. Set force=True to override, but this will be recorded in the event log.

List available meeting templates and their round structures.

Returns: templates: All available templates with round structure details

List meetings for a team, optionally filtered by status.

Start a structured 4-round debate meeting between an Advocate and a Critic.

Debate structure:

• Round 1 (Advocate): Present proposal/position with evidence

• Round 2 (Critic): Challenge risks, flaws, and propose alternatives

• Round 3 (Advocate): Respond to challenges, revise proposal if needed

• Round 4 (Judge): Render verdict with action items

Start a debate-style code review for a specific file or change.

Creates a structured 4-round debate where:

• Advocate defends the current implementation

• Critic challenges the implementation and proposes improvements

• Judge synthesizes findings into consensus conclusions and action items

Update meeting fields (topic, participants, notes).

Use this to add conclusions/notes to a meeting or update its topic. To formally conclude a meeting (mark as concluded), use meeting_conclude instead.

Check which expected participants have spoken in the current round.

Use this after spawning all Agents via dispatch_plan to verify attendance before advancing to the next round or concluding the meeting.

Create a task in a team, waiting for an Agent to pick up and execute.

Rule: Set priority (critical/high/medium/low) and horizon (short/mid/long). Use depends_on for dependencies; the system auto-manages BLOCKED status. Coordinate parallel execution — don't wait for one to complete before starting the next.

Decompose a large task into a parent task + subtasks.

Supports two approaches:

1. Use a built-in template to auto-generate subtasks

2. Manually specify a subtask list

Available templates: web-app, api-service, data-pipeline, library, refactor, bugfix

Create a new task in a project (not bound to a team).

Project-level tasks are attached directly to the project and visible on the project task wall. Suitable for planning-phase tasks not yet assigned to a team.

Query the current status of a task.

Update a task's fields (partial update — only provided fields are changed).

Get intelligent task-Agent matching suggestions.

Analyzes the match between pending unassigned tasks and idle/offline Agents in the team, returning recommended assignments sorted by match_score.

List subtasks of a parent task.

Get the task wall view — categorized by short/mid/long term with intelligent sorting.

Returns a task list sorted by score, helping Leader quickly understand what to do next.

Get project-level task wall — tasks belonging to a project (across all teams).

Unlike taskwall_view (which is team-scoped), this returns tasks from all teams under a project plus standalone project-level tasks.

Read all memo records for a task — read before picking up a task to understand historical progress.

Add a memo record to a task — for tracking progress, recording decisions, marking issues.

Get complete execution timeline for a task.

Returns a unified chronological timeline of all memo records and task lifecycle events, showing who did what, when, and with what result.

Create a new project with a default Phase automatically created.

⚠️ IMPORTANT: Projects are automatically registered by the OS when a CC session starts. You should NOT manually create projects unless the auto-registered project is missing. The root_path MUST match the current CC session's working directory — do NOT create projects pointing to other directories.

List all projects in the system.

Returns: projects: List of all projects with id, name, description, root_path, etc.

Update a project's name, description, or root_path.

Delete a project.

Get a quick project summary: status (active/inactive), teams, top tasks.

Create a new development phase in a project.

Mark current cwd as dismissed for project registration — won't ask again.

List all Phases and their statuses for a project.

Start the company loop — Leader continuous work mode.

After starting, continuously picks up highest-priority tasks. Triggers review discussion every N tasks. When tasks are insufficient, organize meetings to discuss direction; don't create busywork.

Tip: Use /continuous-mode to get the full continuous work protocol, including loop pickup, pause/resume, member management, and detailed behavioral guidelines.

View current company loop status — phase, cycle, completed task count.

Get the next task to execute — sorted by priority x time horizon x readiness.

Pinned and critical tasks are picked up first. short > mid > long priority. BLOCKED tasks auto-unlock when dependencies complete; no manual handling needed.

Advance the loop to the next phase.

Available triggers:

• tasks_planned: Planning done -> Execute

• batch_completed: A batch of tasks completed -> Monitor

• all_tasks_done: All completed -> Review

• issues_found: Issues found -> Return to Execute

• all_clear: All clear -> Review

• new_tasks_added: New tasks added -> Re-plan

• no_more_tasks: No more tasks -> Idle

Pause the loop — preserve current state, can be resumed at any time.

Resume the loop — continue from where it was paused.

Trigger a company loop review — auto-create a review meeting and generate statistics report.

The review meeting contains: summary of tasks completed this cycle, failed task analysis, and next-step suggestions. Leader and team can discuss and produce new to-do tasks in the meeting.

Initialize a pipeline on a task using a lifecycle template.

Writes PipelineState into task.config['pipeline'] and records the first stage in stage_history. Does NOT generate ceremonial subtasks.

Advance the pipeline to the next (or specified) stage.

When force=False, exit conditions are evaluated before advancing. When force=True, exit checks are skipped (Leader override).

Get current pipeline state and recent stage history for a task.

Query team decision log — task assignments, approach selections, Agent scheduling decisions.

List tracked Agent template versions and usage counts.

Shows which template versions (identified by content hash) have been used, when they were first used, and how many times each version was invoked.

Return effectiveness statistics for Agent templates.

Aggregates activity records to compute success rate, average duration, and top failure reasons per template. Also shows how many failure alchemy lessons are associated with each template.

Use this to identify which Agent templates perform well and which need prompt improvement.

Query cross-domain reference edges for an object (who references it / what it references).

Edges are extracted automatically (zero-LLM regex) from task memos and reports: wf_ runs, commit hashes, task UUIDs, [[memory]] links.

Trace the reference neighborhood of an object (undirected fanout, depth <= 2).

Answers questions like "which tasks/reports touched commit 9d8f020" or "what work is connected to run wf_cbad7348".

Search across all OS knowledge: task memos, reports, and tasks.

Three-arm RRF fusion (k=60): BM25 full-text (Chinese bigram native), knowledge-graph fanout (queries containing wf_/commit/uuid IDs pull in everything linked to them), and exact ID-prefix / title match.

Use this to recall past work: "归属铁律怎么修的", "wf_d01f207f", "stderr 盲区", commit hashes, etc.

Save a research/analysis report to the database.

Reports are stored in the database with project isolation — no filesystem permission needed. Reports appear on the Dashboard reports page automatically.

List saved reports, optionally filtered by author, topic, or type.

Returns reports for the current project context, sorted newest-first.

Read the full content of a saved report by ID.

Add a decision item to Leader Briefing for user review.

Use when Leader encounters decisions that require user input: project direction, architecture choices, budget/resource allocation.

List Leader Briefing items. Default shows pending items for user review.

Resolve a Leader Briefing item with user's decision.

Dismiss a Leader Briefing item (no action needed).

Create a scheduled task that triggers automatically on a fixed interval.

List all scheduled tasks, optionally filtered by team.

Pause a scheduled task (set enabled=False).

Permanently delete a scheduled task.

Analyze failed tasks, distill defense rules + training cases + improvement proposals (failure alchemy).

When a task permanently fails (exceeds retry limit), call this tool for deep failure analysis. Automatically generates three learning artifacts saved to team memory:

• Antibody: Defensive rule suggestions to prevent similar failures

• Vaccine: Structured failure case for new Agents to reference and learn from

• Catalyst: System improvement proposals to drive process optimization

Auto-diagnose why a task failed and suggest fixes.

Reads the task's execution trace (memos) to identify the failure point, compares with similar successful tasks in the same team, and returns actionable fix suggestions.

Use this when a task fails or gets stuck to quickly understand root cause without manually reading through all memo records.

Perform What-If analysis on a task — generate multi-approach comparison and recommendation.

During task planning, generates 2-3 alternative approaches with quick scoring comparison:

• Approach A: Best role-match assignment (lowest risk)

• Approach B: Parallel split execution (faster, appears when idle agents >= 2)

• Approach C: History-driven based on experience (appears when team has memory)

Get full execution replay for a task — timeline, checkpoints, stats.

Returns a step-by-step replay of the task execution including:

• timeline: all memo records and lifecycle events in chronological order

• checkpoints: key decision/summary points only

• stats: duration, step count, subtask count, memo type breakdown

Use this to review how a task was executed, understand the decision trail, or audit agent behavior post-completion.

Compare two task executions side by side.

Fetches full replay data for both tasks and produces a diff highlighting:

• step count difference

• checkpoint count difference

• duration difference

• authors unique to each execution vs shared

Useful for comparing a failed run against a successful one, or benchmarking different agent assignments on the same type of task.

Search the memory store in AI Team OS.

Query the team knowledge base — retrieve accumulated experience and lessons learned.

Returns memories with scope=team for this team, including:

• failure_alchemy: Lessons from failure alchemy

• lesson_learned: Manually recorded experiences

• loop_review: Loop review summaries

New Agents should call this tool before joining to get team historical knowledge for quick onboarding.

Add a direction-layer memory — the team's shared, cross-task standing preferences.

方向层 = 低频·高价值密度·跨任务长寿命的偏好/纠正/约束/设计意图。每个派出 的 agent 出生即注入方向层，"全中文""完成即汇报"这类偏好不再靠手抄进 prompt。

写入检验（软门槛）：这条能影响多少未来任务？只影响单个任务的 → 去 task_memo_add（情景层），不要写这里。 方向层的价值在小而准，不在多—— 每作用域有效条目 ≤ 40、单条 ≤ 400 字，超限会被拒绝并提示用 memory_reconcile 先整理。超长内容改写成「触发条件 + 指向权威文件」的指针条目（如 "涉及生产/集群/DB 时遵守只读铁律，详见 ~/.claude/CLAUDE.md"），正文外置。

kind 四类（决定注入截断优先级 constraint>design>directive>preference）：

• constraint（禁令/护栏）：一句话、可机检、终身有效。 如 "所有输出使用中文"、"git 提交绝不自动加 agent 署名"。

• design（价值排序/设计意图）：缺显式指令时的取舍依据。 如 "技术决策偏向质量/简洁/健壮/长期可维护，不看重开发成本"。

• directive（方法论/工作方式）：回答"怎么干"。 如 "完成即按问题→根因→解法→验证汇报，不攒批次"。

• preference（格式偏好）：可选，如 "每句一行便于 diff"。

Invalidate a direction-layer memory — mark it invalid without deleting.

方向层偏好过时/被推翻时显式失效（Zep 失效语义：置 invalid_at 不删除， 保留可审计轨迹）。失效后不再进注入，也默认不出现在 memory_list。

List direction-layer memories — valid entries by default, grouped by kind.

返回当前上下文的方向层条目：global + user 全局条目 + 当前项目的 project 级条目，按 kind 优先级（constraint>design>directive>preference）+ 时间倒序。 这是双 hook 常驻注入的同一数据源；用它审阅"派出的 agent 会继承什么"。

按需整理·粗筛：返回情景层候选组 + 方向层清单 + 蒸馏素材 + 操作说明。

记忆整理 = 会话内按需显式动作（CC 非常驻，无后台整理进程）。本工具只做 确定性粗筛（零 LLM）——OS 无独立 LLM 凭据，判定由你（调用工具的会话内 agent）完成，工具只负责候选粗筛与操作应用（"agent 算、工具存"）。

返回四块（project_id 自动按当前上下文解析）：

• candidate_groups：有效 task_memos 按 scope_path/task 聚簇、簇内 BM25 两两 相似度超阈配对成的候选组（含组内各条全文 + id）。逐组做 LLM 精判： KEEP（都留）/ MERGE（合并）/ INVALIDATE（矛盾失效）/ NOOP（不动）。

• direction_inventory：全部有效方向层条目全文——逐条做陈旧检查（引用的 功能已退役/版本过时/世界已变 → 提 invalidate）。

• promotion_candidates：高频跨任务反复出现的簇，蒸馏为方向层条目的素材 （promote 操作，source_refs 回指源 memo）。

• operation_guide：四操作语义 + reconcile 三守则（只留高频有用 / 指向权威 而非复述 / 重写精简优先）+ 量大开 ultracode 提示。

判完后把确认的操作交给 memory_reconcile_apply 批量应用。

按需整理·应用：批量执行 LLM 精判确认后的操作（确定性，幂等）。

每条操作是一个 dict，按 op 字段分派（未知/缺字段返回 error，不阻断其余）：

• merge：{op:"merge", content:合并后新内容, memo_ids:[被并各条], memo_type?:"summary", scope_path?} —— 建新 memo，把被并各条置 invalid、 invalidated_by 指向新条（Zep 失效语义不删除）。

• invalidate：{op:"invalidate", memo_ids:[...]} —— 逐条失效（矛盾/被推翻）。

• score：{op:"score", memo_id, quality_score:1-10, reason} —— 补质量分， reason 入 meta。

• promote：{op:"promote", content, kind:constraint/design/directive/preference, scope?:"project"/global/user, source_refs?:[源 memo id]} —— 蒸馏提升为方向层 条目；红线照常生效（单条 ≤400 字、每桶有效 ≤40 条，超限该条返回 error）。

• keep / noop：不动（可省略）。

幂等：对已失效条目重复 invalidate/merge 返回 noop 不报错。应用后自动刷新 项目 last_reconcile_at（量阈软提示的基线）。

Record an agent execution pattern (success or failure) for future learning.

Stores the pattern in the global execution pattern memory so future agents can benefit from this experience when tackling similar tasks.

Search historical execution patterns similar to a task description.

Uses BM25 retrieval to find relevant success/failure patterns recorded by agents in previous tasks. Use this before starting a complex task to benefit from past experience.

Get the current active OS context — active project, active team, member list, loop status.

This is the infrastructure for all simplified operations. A single call returns the complete context of the current working environment, allowing Leader or other tools to auto-fill parameters like project_id, team_id, etc.

Returns: Context dict containing project / team / agents / loop

Check the health status of the AI Team OS API service.

Verifies the API service is running normally by accessing the team list endpoint.

Returns: Health status info including API reachability and team count

Restart the AI Team OS FastAPI process safely (standardized restart flow).

Use this after backend code changes to pick up the new version without manually killing processes. The flow has three safety guards:

1. Busy-agent guard — refuses to restart while any agent is working (status=busy) unless force=True.

2. Port-pin guard — only ever restarts on the ORIGINAL port (default 8000, read from api_port.txt). If that port is held by an unrelated process it aborts rather than drifting to a random port.

3. Dead-before-spawn guard — waits until the old process has fully exited and released the port before spawning the new one; never spawns on a timeout.

If the API is already down, steps 2-4 are skipped and this becomes a plain "start" of the API on its configured port.

Report an issue to the team. Issues are created as high-priority tasks, auto-tagged as issue type.

Severity maps to task priority: critical->critical, high->high, medium->high, low->medium.

Mark an Issue as resolved with a resolution description.

Updates the Issue status to resolved and records the resolution. The corresponding task is also marked as completed.

List recent events in the system.

Find ecosystem skills/plugins using a 3-layer progressive loading system.

Layer 1 (quick recommend): Describe your task and get top 3-5 matching skills with one-line descriptions and install commands. Layer 2 (category browse): Browse all skills grouped by category (memory / code-quality / frontend / security / dev-workflow / etc.). Layer 3 (full detail): Get complete documentation for a single skill including features, OS complement relationship, and variants.

Send a notification to the configured Slack webhook.

Requires SLACK_WEBHOOK_URL to be configured via PUT /api/settings/webhook.

List available ecosystem integration recipes for combining AI Team OS with external tools.

Recipes describe how to integrate external MCP servers (GitHub, Slack, Linear, etc.) with AI Team OS workflows. Each recipe includes: recommended MCP server, install config, and concrete collaboration scenarios with AI Team OS tools.

Send a message to another project (or broadcast to all).

Messages are stored in the shared global DB so any project can read them. Requires PROJECT_DIR env var (set automatically by Claude Code via CLAUDE_PROJECT_DIR).

Read the cross-project message inbox for the current project.

Returns direct messages sent to this project plus any broadcasts. Requires PROJECT_DIR env var (set automatically by Claude Code via CLAUDE_PROJECT_DIR).

Get model governance state: available models (auto-discovered from local CC transcripts — the models you actually used) and the current default startup model (~/.claude/settings.json "model" key).

Set the default startup model for new CC sessions (writes the "model" key in ~/.claude/settings.json; empty string removes the key, restoring CC's own default). Takes effect on NEW sessions.

Declare exclusive edit intent on a file to prevent concurrent modifications.

Call this before editing a shared file (types.py, models.py, etc.). If another agent already holds the lock, the call fails with details about who holds it and how long until it expires.

Release the file lock after editing is complete.

Call this immediately after you finish editing a file you locked with file_lock_acquire. This allows other agents to proceed without waiting for the TTL to expire.

List all currently active file locks held by agents.

Useful for team-lead to inspect the workspace state or diagnose potential conflicts before dispatching concurrent agents.

Returns: {"locks": [{"path": ..., "agent": ..., "expires_in": ...}, ...], "count": N}