API Exports Index¶
Deep links for ultra-granular browsing, with brief descriptions of what, when, and how. For HTTP endpoints, use the REST API Overview.
Decorators¶
-
Decorators Overview¶
- Why: Fastest way to convert Python functions into discoverable, callable tools/agents.
- What: Schema inferred from type hints; registered into the catalog for planners, REST, and CLI.
- When: Use
mcp_toolfor public tools,toolfor lightweight utilities,a2a_agentfor delegatable agents. - tool — Register a generic tool from a Python function.
- mcp_tool — Register an MCP-style tool with structured parameters for discovery.
- a2a_agent — Wrap an agent function to delegate via the A2A client.
Templates¶
-
Templates Overview¶
- Why: Maximum control over tool definitions, parameters, validation, and lifecycle.
- What: Class-based tools with explicit schemas and custom behaviors.
- When: Use when decorators are too limiting or for sandbox/code-exec scenarios.
- BaseTemplate — Abstract base for custom template behaviors.
- FunctionToolTemplate — Define function-style tools with explicit schemas.
- MCPToolTemplate — MCP-flavored template with richer discovery metadata.
- CodeExecToolTemplate — Execute user-provided code with sandbox controls.
- AgentTemplate — Define agent-style tools (evaluators, classifiers).
YAML Loaders¶
-
YAML Loaders Overview¶
- Why: Configuration-driven registration for repeatable deployments and non-Python authors.
- What: Load single files or directories of YAML; validated schemas with clear error types.
- When: Use in DevOps pipelines or to share tools without Python code changes.
- load_tools_from_yaml — Load a single YAML tool definition.
- load_tools_from_directory — Bulk load YAML tool definitions from a directory.
- YAMLLoaderError — General loading error (IO/parse).
- YAMLValidationError — Schema validation error.
- WorkerResolutionError — Referenced backend/worker cannot be resolved.
Skill Bridge¶
-
Skill Bridge Overview¶
- Why: Persist and share tools as reusable skills across teams/environments.
- What: Save, load, sync skills; list tools backed by skills.
- When: Promote approved tools to a registry for standardized reuse.
- save_tool_as_skill — Persist a tool as a reusable skill package.
- load_tool_from_skill — Load a skill-backed tool for reuse.
- get_tool_skill — Retrieve a tool’s associated skill metadata.
- sync_tool_with_skill — Synchronize tool changes to its skill package.
- get_skill_backed_tools — List tools backed by skills.
Discovery¶
-
Discovery Overview¶
- Why: Find and browse the right tools for planners, UIs, and workflows.
- What: List all tools, keyword/domain search, semantic matching, and progressive browsing.
- When: Build catalogs, surface tools to users, or feed planners.
- get_available_tools — Return all registered tools.
- search_tools — Filter tools by domain/keywords.
- get_tool_info — Detailed definition for a tool by name.
- list_tools_by_domain — List tools within a given domain.
- semantic_search_tools — Natural language matching against tool descriptions.
- browse_tools — Progressive detail browsing suitable for CLI/UX.
Plugins¶
-
Plugins Overview¶
- Why: Extend ToolWeaver with new discovery sources, integrations, and custom behaviors.
- What: Register, list, get, unregister, and auto-discover plugins.
- When: Add bespoke capabilities without modifying core.
- register_plugin — Register a plugin instance.
- unregister_plugin — Remove a plugin by name.
- get_plugin — Retrieve a plugin instance.
- list_plugins — List all registered plugins.
- discover_plugins — Scan and auto-register discoverable plugins.
Configuration¶
-
Configuration Overview¶
- Why: Control runtime behavior via environment variables and defaults.
- What: Read, reset, and validate configuration; typical fields include cache path, Redis URL, skill path, and logging.
- When: Setup environment, troubleshoot issues, and standardize deployments.
- get_config — Read current runtime configuration.
- reset_config — Reset cached configuration state.
- validate_config — Validate configuration and return warnings/errors.
Logging¶
-
Logging Overview¶
- Why: Provide observability with safety (secrets never leak into logs).
- What: Module-specific loggers, global log level, and a secrets redactor installed via debug mode.
- When: Standardize logging across apps and CI while protecting credentials.
- get_logger — Get a module-specific logger with safe defaults.
- set_log_level — Set global log level.
- enable_debug_mode — Install secrets redactor and enable safe logging.
A2A Client¶
-
A2A Client Overview¶
- Why: Delegate work to other agents and compose agent networks.
- What: Capabilities, delegation requests/responses, and
A2AClientfor execution. - When: Orchestrate multi-agent workflows and federated capabilities.
- AgentCapability — Describe a capability an agent provides.
- AgentDelegationRequest — Request payload to delegate to an agent.
- AgentDelegationResponse — Response payload from delegated agent.
- A2AClient — Client to delegate requests to agents.
Version & Security¶
-
Version & Security Overview¶
- Why: Track package version and leverage built-in safety features.
- What:
__version__string and an auto-installed secrets redactor on import. - When: Display version in UIs/logs and ensure safe logging defaults.
- version — Package version string.
- secrets redactor — Auto-installed redactor to prevent secrets appearing in logs.