ms-agent

# ms-agent Capability Gateway Skill This is a **single skill package** (`name: ms-agent` in frontmatter) with **9 capability reference modules** under `references/`. It connects you to the **ms-agent Capability Gateway** — a unified interface to ms-agent's projects, components, and atomic tools, exposed as MCP tools. ### Skill identity (`skill_id`) When loaded by ms-agent's own SkillLoader, `skill_id` equals the directory name: | How loaded | Directory | `skill_id` | |---|---|---| | Host install (`examples/capability/*/install_skill.sh`) | host `skills/ms-agent/` | `ms-agent` | | ms-agent itself (`scripts/install_into_ms_agent.sh`) | `$MS_AGENT_HOME/skills/ms-agent/` (default `~/.ms_agent/skills/ms-agent`) | `ms-agent` | | Wheel / `pip install ms-agent` | `ms_agent/skills/ms-agent/` (built-in) | `ms-agent` | | Repo folder loaded directly | `ms-agent-skills/` | `ms-agent-skills` | The frontmatter `name: ms-agent` is the package identifier; `skill_id` follows the filesystem directory name at load time. ## Setup Verify ms-agent is installed: ```bash python scripts/check_ms_agent.py python scripts/validate_capability_index.py ``` To make this skill discoverable by ms-agent's own SkillCatalog: ```bash ./scripts/install_into_ms_agent.sh ``` The MCP server must be configured in your agent's config. Pick the section that matches your agent host. ### nanobot config.json ```json { "tools": { "mcpServers": { "ms-agent": { "command": "python3", "args": ["-m", "ms_agent.capabilities.mcp_server"], "env": {"PYTHONPATH": "/path/to/ms-agent"}, "toolTimeout": 300, "enabledTools": ["*"] } } } } ``` ### OpenClaw openclaw.json ```json { "mcp": { "servers": { "ms-agent": { "command": "/absolute/path/to/python", "args": ["-m", "ms_agent.capabilities.mcp_server"], "env": {"PYTHONPATH": "/path/to/ms-agent"} } } } } ``` ### Hermes Agent config.yaml ```yaml mcp_servers: ms-agent: command: "python3" args: ["-m", "ms_agent.capabilities.mcp_server"] env: PYTHONPATH: "/path/to/ms-agent" ``` Once configured, all 30 capabilities below are available as MCP tools. The tool prefix depends on your agent host: | Agent | Tool name format | Example | |-------|-----------------|---------| | nanobot | `mcp_ms-agent_<tool>` | `mcp_ms-agent_web_search` | | OpenClaw | `ms-agent_<tool>` | `ms-agent_web_search` | | Hermes | `mcp_ms-agent_<tool>` | `mcp_ms-agent_web_search` | | Cursor / Claude Desktop | `<tool>` (no prefix) | `web_search` | ## Capability Index ### Deep Research | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `submit_research_task` | async submit | seconds | [deep-research.md](references/deep-research.md) | | `check_research_progress` | async poll | seconds | [deep-research.md](references/deep-research.md) | | `get_research_report` | async result | seconds | [deep-research.md](references/deep-research.md) | | `deep_research` | sync (blocks) | hours (20–60 min wall) | [deep-research.md](references/deep-research.md) | ### Document Research | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `submit_doc_research_task` | async submit | seconds | [doc-research.md](references/doc-research.md) | | `check_doc_research_progress` | async poll | seconds | [doc-research.md](references/doc-research.md) | | `get_doc_research_report` | async result | seconds | [doc-research.md](references/doc-research.md) | | `doc_research` | sync (blocks) | minutes (5–20 min wall) | [doc-research.md](references/doc-research.md) | ### Financial Research | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `submit_fin_research_task` | async submit | seconds | [fin-research.md](references/fin-research.md) | | `check_fin_research_progress` | async poll | seconds | [fin-research.md](references/fin-research.md) | | `get_fin_research_report` | async result | seconds | [fin-research.md](references/fin-research.md) | | `fin_research` | sync (blocks) | hours (20–60 min wall) | [fin-research.md](references/fin-research.md) | ### Web Search | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `web_search` | sync | seconds | [web-search.md](references/web-search.md) | ### Code Generation | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `submit_code_genesis_task` | async submit | seconds | [code-genesis.md](references/code-genesis.md) | | `check_code_genesis_progress` | async poll | seconds | [code-genesis.md](references/code-genesis.md) | | `get_code_genesis_result` | async result | seconds | [code-genesis.md](references/code-genesis.md) | | `code_genesis` | sync (blocks) | hours (10–30 min wall) | [code-genesis.md](references/code-genesis.md) | ### Video Generation | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `submit_video_generation_task` | async submit | seconds | [singularity-cinema.md](references/singularity-cinema.md) | | `check_video_generation_progress` | async poll | seconds | [singularity-cinema.md](references/singularity-cinema.md) | | `get_video_generation_result` | async result | seconds | [singularity-cinema.md](references/singularity-cinema.md) | | `video_generation` | sync (blocks) | hours (~20 min wall) | [singularity-cinema.md](references/singularity-cinema.md) | ### Agent Delegation | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `delegate_task` | sync (blocks) | minutes | [agent-delegate.md](references/agent-delegate.md) | | `submit_agent_task` | async submit | seconds | [agent-delegate.md](references/agent-delegate.md) | | `check_agent_task` | async poll | seconds | [agent-delegate.md](references/agent-delegate.md) | | `get_agent_result` | async result | seconds | [agent-delegate.md](references/agent-delegate.md) | | `cancel_agent_task` | async cancel | seconds | [agent-delegate.md](references/agent-delegate.md) | ### Code Validation | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `lsp_check_directory` | full scan | minutes | [lsp-code-server.md](references/lsp-code-server.md) | | `lsp_update_and_check` | incremental | seconds | [lsp-code-server.md](references/lsp-code-server.md) | ### File Editing | Tool | Type | Duration | Reference | |------|------|----------|-----------| | `replace_file_contents` | content-match | seconds | [filesystem-tools.md](references/filesystem-tools.md) | | `replace_file_lines` | line-range | seconds | [filesystem-tools.md](references/filesystem-tools.md) | ## Quick Decision Guide ``` User wants to... │ ├── Research a topic in depth (20-60 min) │ └── submit_research_task → check_research_progress → get_research_report │ ├── Analyze documents or URLs (5-20 min) │ └── submit_doc_research_task → check_doc_research_progress → get_doc_research_report │ ├── Financial analysis with data & sentiment (20-60 min) │ └── submit_fin_research_task → check_fin_research_progress → get_fin_research_report │ ├── Search the web for quick info │ └── web_search(query="...", engine_type="arxiv|exa|serpapi|tavily") │ ├── Generate a software project from requirements (10-30 min) │ └── submit_code_genesis_task → check_code_genesis_progress → get_code_genesis_result │ ├── Generate a short video (~ 20 min) │ └── submit_video_generation_task → check_video_generation_progress → get_video_generation_result │ ├── Delegate a complex multi-step task to an AI agent │ ├── Short task (< 5 minutes) → delegate_task(query="...") │ └── Long task → submit_agent_task → check_agent_task → get_agent_result │ ├── Validate code for errors │ ├── Full project → lsp_check_directory(directory="src/", language="typescript") │ └── Single file → lsp_update_and_check(file_path="...", content="...", language="...") │ └── Edit a file precisely ├── Know exact text → replace_file_contents (concurrent-safe) └── Know line numbers → replace_file_lines ``` ## Async Pattern All project-level capabilities (deep research, doc research, financial research, code generation, video generation, agent delegation) support an async submit/check/get pattern. This does NOT block the agent — continue handling other messages while the task runs in the background. ``` 1. submit_*_task(...) → returns task_id immediately 2. check_*_progress/task(task_id) → poll status (repeat every few minutes) 3. get_*_result/report(task_id) → retrieve final result when completed 4. cancel_*_task(task_id) → cancel if no longer needed (agent delegation only) ``` **Warning — in-process task tracking:** `AsyncTaskManager` is **in-process only**. Task IDs are lost if the MCP server process restarts, and orphaned subprocesses may remain. Clients must keep the **same MCP server process** for the full submit → check → get lifecycle. Read the reference files for detailed SOP workflows for each capability.