菜单
Use this skill for creating AI agents, setting up knowledge bases (RAG), configuring LLM models, building tools (API/MCP), managing conversations, streaming chat, or querying LLMs directly on SELISE Blocks. Note: this API uses snake_case throughout.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use this skill for deploying code, managing builds, monitoring services, and configuring runtime environments on SELISE Blocks.
Use this skill for managing scheduled tasks, webhooks, configuration settings, and other utility operations on SELISE Blocks.
Use this skill for creating, managing, and executing automated workflows with webhook/email triggers, AI agent nodes, HTTP request nodes, and email sending nodes on SELISE Blocks.
Use this skill for sending emails (template-based or ad-hoc), managing in-app notifications (send, read, mark-read), and creating/managing email templates on SELISE Blocks.
Use this skill for defining data schemas, querying/mutating data via GraphQL, uploading files (S3/DMS), managing access policies, setting field validations, migrating schemas, or configuring data sources on SELISE Blocks.
Use this skill for any request involving login, MFA setup, user creation, role/permission management, organization switching, SSO/OIDC configuration, session handling, password recovery, CAPTCHA, or access control on SELISE Blocks. Maps natural language intents to the correct flow or action and enforces security best practices.
| name | ai-services |
| description | Use this skill for creating AI agents, setting up knowledge bases (RAG), configuring LLM models, building tools (API/MCP), managing conversations, streaming chat, or querying LLMs directly on SELISE Blocks. Note: this API uses snake_case throughout. |
| user-invocable | false |
| blocks-version | 1.0.3 |
Handles all AI agent management, knowledge base ingestion, tool configuration, model setup, and conversation operations for SELISE Blocks via the blocksai-api v1 API.
Important: This API uses snake_case (Python/FastAPI style) throughout. All request and response fields use
snake_case— e.g.,project_key,agent_id,is_success,kb_ids,tool_ids. Do NOT use camelCase when constructing requests or reading responses.
Must run get-token before any action in a session.
Example prompts that should route here:
Before executing any action or flow from this skill, read ../core/execution-context.md for the required supporting files, load order, and cross-domain orchestration rules.
All endpoints are prefixed with: $VITE_API_BASE_URL/blocksai-api/v1
Use this table to route user requests. Check flows/ first — if a flow covers the request, use it. For single-action requests, go directly to the action.
| User wants to... | Use |
|---|---|
| Create an AI agent from scratch | flows/create-agent-flow.md |
| Send a prompt to an LLM directly without an agent | flows/query-lmt-flow.md |
| Stream a response from a language model | flows/query-lmt-flow.md |
| Configure an agent's model, KB, tools | flows/create-agent-flow.md |
| Publish an agent to marketplace | actions/publish-agent.md |
| Set up a knowledge base with files/text/links | flows/setup-knowledge-base.md |
| Upload files to a knowledge base | actions/upload-kb-file.md |
| Add text content to a knowledge base | actions/ingest-kb-text.md |
| Add Q&A pairs to a knowledge base | actions/ingest-kb-qa.md |
| Crawl a URL into a knowledge base | actions/ingest-kb-link.md |
| Test knowledge base retrieval | actions/test-kb-retrieval.md |
| Chat with an AI agent | flows/chat-flow.md |
| Start a conversation session | actions/initiate-conversation.md |
| Send a message to an agent | actions/chat-agent.md |
| Stream a response from an agent | actions/chat-sse.md |
| Query the LLM directly (no agent) | actions/query-lmt.md |
| Stream a direct LLM query | actions/stream-query-lmt.md |
| Add an AI model config | flows/manage-models.md |
| List available models | actions/get-models.md |
| Validate a model's API key | actions/validate-model.md |
| Create an API tool for an agent | actions/create-api-tool.md |
| Create an MCP server tool | actions/create-mcp-tool.md |
| Test a tool action | actions/test-tool-action.md |
| List all tools | actions/get-tools.md |
| Delete a tool | actions/delete-tool.md |
| List agents | actions/get-agents.md |
| Get agent details | actions/get-agent.md |
| Update agent name/description/persona | actions/update-agent-persona.md |
| Update agent AI configuration | actions/update-agent-ai-config.md |
| Enable or disable an agent | actions/change-agent-status.md |
| Delete an agent | actions/delete-agent.md |
| List conversation sessions | actions/get-conversations.md |
| Delete a conversation session | actions/delete-conversation.md |
| Flow | File | Description |
|---|---|---|
| create-agent-flow | flows/create-agent-flow.md | Create agent → configure AI settings → add KB → add tools → publish |
| setup-knowledge-base | flows/setup-knowledge-base.md | Create folder → upload files/text/links → test retrieval → attach to agent |
| chat-flow | flows/chat-flow.md | Initiate conversation → send messages → handle streaming → manage session |
| manage-models | flows/manage-models.md | List providers → get available models → create model config → validate API key |
| Action | File | Description |
|---|---|---|
| create-agent | actions/create-agent.md | Create a new AI agent from a description |
| update-agent-persona | actions/update-agent-persona.md | Update agent name, description, and persona |
| update-agent-ai-config | actions/update-agent-ai-config.md | Update model, temperature, system prompt, KB and tool attachments |
| change-agent-status | actions/change-agent-status.md | Enable, disable, or archive an agent |
| delete-agent | actions/delete-agent.md | Permanently delete an agent |
| get-agents | actions/get-agents.md | List agents with filter and pagination |
| get-agent | actions/get-agent.md | Get a single agent's full details |
| publish-agent | actions/publish-agent.md | Publish an agent to the marketplace |
| Action | File | Description |
|---|---|---|
| upload-kb-file | actions/upload-kb-file.md | Upload and index a file (PDF, DOCX, TXT, etc.) |
| ingest-kb-text | actions/ingest-kb-text.md | Ingest raw text content into a KB folder |
| ingest-kb-qa | actions/ingest-kb-qa.md | Ingest Q&A pairs into a KB folder |
| ingest-kb-link | actions/ingest-kb-link.md | Crawl and index a URL into a KB folder |
| create-kb-folder | actions/create-kb-folder.md | Create a KB folder with embedding configuration |
| delete-kb | actions/delete-kb.md | Delete a knowledge base entry |
| test-kb-retrieval | actions/test-kb-retrieval.md | Test retrieval quality with a query against an agent |
| Action | File | Description |
|---|---|---|
| create-api-tool | actions/create-api-tool.md | Create or update an API tool with actions |
| create-mcp-tool | actions/create-mcp-tool.md | Create an MCP server tool |
| test-tool-action | actions/test-tool-action.md | Test a specific action on a tool |
| get-tools | actions/get-tools.md | List all tools with optional filtering |
| delete-tool | actions/delete-tool.md | Delete a tool permanently |
| Action | File | Description |
|---|---|---|
| create-model | actions/create-model.md | Create a new AI model configuration |
| get-models | actions/get-models.md | List all configured AI models |
| get-model | actions/get-model.md | Get a single model's details |
| validate-model | actions/validate-model.md | Validate a model's API key and configuration |
| get-provider-models | actions/get-provider-models.md | Get available models from a provider (seed data) |
| Action | File | Description |
|---|---|---|
| initiate-conversation | actions/initiate-conversation.md | Start or reconnect to a chat session for an agent |
| get-conversations | actions/get-conversations.md | List all chat sessions with filter and pagination |
| delete-conversation | actions/delete-conversation.md | Delete a chat session permanently |
| Action | File | Description |
|---|---|---|
| query-lmt | actions/query-lmt.md | Query the language model directly (non-streaming) |
| stream-query-lmt | actions/stream-query-lmt.md | Stream a language model query response |
| chat-agent | actions/chat-agent.md | Send a message to an agent workspace |
| chat-sse | actions/chat-sse.md | Send a message via SSE chat endpoint (Server-Sent Events) |
