AgentMarketMCP / SKILL 资产档案馆

目录 / Danube

MCP 需 API Key 未评级 已上架

Danube

The App Store for Agent Tools

该来源不提供完整文件导出（国内平台多为平台内托管），仅存元数据与原链

接入信息

传输形态
http
鉴权方式
需 API Key（需要配置：danube-api-key）
端点
https://danube--preston.run.tools
该服务需要凭证，请按官方文档申请后替换占位符
{
  "mcpServers": {
    "Danube": {
      "headers": {
        "Authorization": "Bearer \u003cYOUR_KEY\u003e"
      },
      "url": "https://danube--preston.run.tools"
    }
  }
}

能力清单

工具说明
list_servicesList available services. Optionally filter by query. Parameters: - query: Optional search query to filter services by name or description - limit: Maximum number of services to return (default 10) Returns a list of available service providers with their tool counts.
search_toolsSearch for tools by query (matches name or description). Parameters: - query: Search query that matches against tool names and descriptions - service_id: Optional filter to only search within a specific service - limit: Maximum number of results to return (default 10) Uses semantic search to find the most relevant tools.
get_service_toolsGet all tools for a specific service. Parameters: - service_id: The ID of the service to get tools for - limit: Maximum number of tools to return (default 50) Returns: - tools: List of available tools for the service - needs_configuration: True if the service requires credentials that aren't configured - skipped: True if service was skipped due to user's configured_only policy - configuration_required: Details about how to configure credentials (when needs_configuration=True) If the service requires credentials that haven't been configured: - Returns needs_configuration=True with instructions on how to configure - The configuration_required field includes service_name, message, and configuration_url
execute_toolExecute a tool by ID or name with parameters. Parameters: - tool_id: The UUID of the tool to execute (preferred, faster) - tool_name: The name of the tool (e.g., "Hacker News - Get Top Stories") - parameters: Dictionary of parameters to pass to the tool - max_response_chars: Optional maximum characters in the response (default: 50000, max: 500000). Use this to request more data from tools that return large structured responses (e.g. flight search). Provide either tool_id OR tool_name. If both are provided, tool_id takes precedence. Returns the result of the tool execution.
get_wallet_balanceGet your current wallet balance. Returns your available credit balance in cents and formatted as currency. Use this to check if you have sufficient funds before executing paid tools. Returns: - balance_cents: Your balance in cents (e.g., 1500 = $15.00) - formatted: Human-readable balance (e.g., "$15.00") - has_wallet: Whether you have a wallet set up
get_spending_limitsGet your current USDC spending limits. Returns your per-call and daily spending limits for USDC tool payments. The platform enforces a hard cap of $5.00 per call (5,000,000 atomic units). Returns: - max_per_call_atomic: Maximum USDC per tool call in atomic units (6 decimals, 1 USDC = 1,000,000) - max_per_call_usdc: Same amount in USDC (e.g., 5.0) - daily_limit_atomic: Daily spending limit in atomic units (null = no daily limit) - daily_limit_usdc: Same in USDC (null = no daily limit)
update_spending_limitsUpdate your USDC spending limits. Controls how much can be spent on paid tool executions. The platform enforces a hard cap of $5.00 per call regardless of your setting. Parameters: - max_per_call_usdc: Maximum USDC per tool call (0.00 to 5.00). Default is 5.00. - daily_limit_usdc: Optional daily spending cap in USDC. Set to 0 to remove the daily limit. Returns the updated spending limits.
store_credentialStore an API key or bearer token credential for a service. Use this after a tool execution returns an auth_required error. Ask the user for their API key for the service, then call this tool to save it. Parameters: - service_id: The service ID that needs credentials (from the auth_required error) - credential_type: Type of credential - "bearer" or "api_key" (default: "bearer") - credential_value: The API key or token value provided by the user Note: This only works for API key/bearer credentials. OAuth credentials must be configured through the Danube dashboard in a browser. Returns success/failure status.
search_skillsSearch for skills by query using semantic search. Skills are reusable instructions that teach AI agents how to perform specific tasks. They contain step-by-step guides, scripts, and reference materials. Parameters: - query: Search query that matches against skill names and descriptions - limit: Maximum number of results to return (default 10) Returns a list of skills matching the query.
get_skillGet a skill by ID or name with full content. Returns the complete skill including: - SKILL.md content with instructions - Scripts (executable code) - References (additional documentation) - Assets (templates, resources) Parameters: - skill_id: The UUID of the skill (preferred, faster) - skill_name: The name of the skill (e.g., "pdf-processing") Provide either skill_id OR skill_name. If both are provided, skill_id takes precedence. Returns the skill with all file contents for the agent to use.
create_skillCreate a new skill. Skills are reusable instructions that teach AI agents how to perform specific tasks. They contain SKILL.md content with step-by-step guides, plus optional scripts and references. Parameters: - name: Skill name (e.g. "pdf-processing") - skill_md_content: The SKILL.md markdown content with instructions. Should include YAML frontmatter with description. - scripts: Optional list of executable script files, each with {name, content} - reference_files: Optional list of reference doc files, each with {name, content} - assets: Optional list of asset files, each with {name, content} - visibility: "private" (default). Public skills require the /skill-submissions review flow. - service_id: Optional service UUID to associate the skill with Returns the created skill with all content.
update_skillUpdate an existing skill. Only the owner can update. Parameters: - skill_id: The UUID of the skill to update - name: New skill name (optional) - skill_md_content: New SKILL.md content (optional) - scripts: New list of script files (optional) - reference_files: New list of reference files (optional) - assets: New list of asset files (optional) Returns the updated skill.
delete_skillDelete a skill. Only the owner can delete. Parameters: - skill_id: The UUID of the skill to delete Returns a confirmation message.
register_agentRegister a new autonomous agent on Danube. Creates a user profile, API key, and USDC wallet in one call. The API key is returned only once — save it immediately. No authentication required. Rate limited to 5 requests per minute. Parameters: - name: Display name for the agent (e.g. "my-research-bot") - operator_email: Contact email for the agent's operator Returns: - agent_id: Unique agent identifier - api_key: Danube API key (shown only once — save it!) - wallet_id: USDC wallet identifier - balance_usdc: Current USDC balance (starts at 0) - deposit_address: Platform wallet address for USDC deposits (Base network) - registration_url: URL to the agent's ERC-8004 registration file
get_agent_infoGet the current agent's profile and wallet balance. Returns agent details including USDC wallet balance and deposit address. Only works for agent accounts (not human user accounts). Returns: - agent_id: Agent identifier - name: Agent display name - agent_type: "autonomous" - operator_email: Operator contact email - wallet: {id, balance_usdc, balance_atomic} - deposit_address: Platform wallet address for USDC deposits - registration_url: URL to the agent's registration file - created_at: Registration timestamp
fund_agent_walletFund the agent's wallet for tool execution payments. Two methods available: - "card_checkout": Creates a Stripe checkout URL. The operator opens the URL in a browser to complete the card payment. Requires amount_cents parameter. - "crypto": Returns the platform USDC deposit address on Base network. Send USDC to this address, then confirm via the confirm-deposit endpoint. Parameters: - method: "card_checkout" or "crypto" - amount_cents: Amount in cents (required for card_checkout, min 100 = $1.00) Returns: - wallet_id: Wallet identifier - method: The funding method used - checkout_url: Stripe checkout URL (card_checkout only) - deposit_address: USDC deposit address (crypto only) - network: Blockchain network (crypto only, e.g. "eip155:8453" for Base) - usdc_contract: USDC contract address (crypto only) - balance_usdc: Current USDC balance
list_workflowsList public workflows, optionally filtered by search query. Workflows are multi-tool orchestrations that chain several Danube tools together with data flowing between steps using template syntax. Parameters: - query: Optional search query to filter by name or description - limit: Maximum number of results (default 10) Returns a list of public workflows with name, description, step count, and tags.
execute_workflowExecute a multi-tool workflow. Runs all steps in order, passing data between them via template syntax. Each step executes a Danube tool with resolved inputs. Parameters: - workflow_id: The UUID of the workflow to execute - inputs: Dictionary of input values for the workflow (referenced as {{inputs.field}} in steps) Returns the full execution result with per-step outcomes.
get_workflow_executionGet the result of a workflow execution. Parameters: - execution_id: The UUID of the execution to retrieve Returns the full execution result including status, per-step results, and timing.
create_workflowCreate a new multi-tool workflow. Supports two modes: - "template": Rigid step-by-step execution with template variable substitution. - "instructions": AI-orchestrated execution using natural language instructions. Parameters: - name: Workflow name - steps: (template mode) List of step definitions, each with: - step_number: Order of execution (starting from 1) - tool_id: UUID of the Danube tool to call - tool_name: Optional display name - description: Optional step description - input_mapping: Dict mapping parameter names to values or templates like {{steps.1.result.field}} or {{inputs.field}} - description: Optional workflow description - visibility: "private" (default) or "public" - tags: Optional list of tags for discoverability - instructions: (instructions mode) Natural language instructions describing what the workflow should do - allowed_tool_ids: (instructions mode) List of tool UUIDs the workflow is allowed to call - input_schema: (instructions mode) List of input definitions, each with "name" and "description" - execution_mode: "template" (default) or "instructions" Returns the created workflow.
update_workflowUpdate an existing workflow. Only the owner can update. Parameters: - workflow_id: The UUID of the workflow to update - name: New workflow name (optional) - description: New description (optional) - steps: New list of step definitions (optional) - visibility: New visibility - "private" or "public" (optional) - tags: New list of tags (optional) Returns the updated workflow.
delete_workflowDelete a workflow. Only the owner can delete. Parameters: - workflow_id: The UUID of the workflow to delete Returns a confirmation message.
batch_execute_toolsExecute multiple tools in a single batch request. Runs up to 10 tool calls concurrently and returns all results. Individual failures don't fail the batch — each call returns independently. Parameters: - calls: List of tool calls, each with: - tool_id: UUID of the tool to execute - tool_input: Dict of parameters for the tool (optional, defaults to {}) Returns a list of results, each with tool_id, success, result, and error fields.
submit_ratingRate a tool (1-5 stars) with optional comment. Updates existing rating if you've already rated this tool. Parameters: - tool_id: The UUID of the tool to rate - rating: Rating from 1 to 5 - comment: Optional text comment about the tool Returns the saved rating.
get_my_ratingGet your rating for a specific tool. Returns your previously submitted rating (1-5) and comment, or null if you haven't rated this tool yet. Parameters: - tool_id: The UUID of the tool to check your rating for Returns your rating and comment, or null values if not yet rated.
get_tool_ratingsGet aggregated ratings for a tool. Returns the average star rating and total number of ratings for a tool. This is a public endpoint — no authentication required. Parameters: - tool_id: The UUID of the tool to get ratings for Returns average_rating (1.0-5.0) and total_ratings count.
report_toolReport a broken or degraded tool. Notifies the tool publisher so they can investigate and fix the issue. Parameters: - tool_id: The UUID of the tool to report - reason: One of: broken, degraded, incorrect_output, timeout, other - description: Optional details about the issue Returns the created report.
get_recommendationsGet tool recommendations based on co-usage patterns. Optionally pass a tool_id to find tools frequently used together with it. Without a tool_id, returns popular tools or personalized recommendations. Parameters: - tool_id: Optional UUID of a tool to find related tools for - limit: Maximum number of recommendations (default 5) Returns a list of recommended tools with reasons.
纠错与举报（发现条目失效、署名有误或涉及侵权？）
提交举报 / 纠错

侵权举报经核验成立后，我们会即时下线该条目并删除已存的内容副本。