Menu
Conversation data model — ConversationContext, Message, ToolCall/ToolResult, polymorphic content. Use when reading/writing context, adding a new content type, or debugging tool-call pairing.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Router for any non-trivial echo-sdk task. Maps user intent to the right module skill and diagram. Invoke FIRST for any change to src/echo/ before reaching for a specific module skill.
Recipe for adding a new provider (LLM, transcriber, eval, prompt). Use whenever adding Cohere, Mistral, Deepgram, Phoenix, Promptlayer, etc. — anything that plugs into an existing get_*() factory.
End-to-end recipe for defining a Skill bundle and wiring it into a GenericAgent with the right activation mode and tool registry. Use when extending agent capability via swappable skills rather than baking tools in.
Building agents with BaseAgent / GenericAgent, including skill registration, activation modes (llm vs manual), and system-prompt composition. Use when editing src/echo/agents/ or building a new agent subclass.
Audio transcription provider abstraction (Gemini, EkaCare). Use when transcribing audio, adding a new transcriber, or shaping AudioInput.
Postgres support — asyncpg client, schema-aware binder, PgQueryTool surface. Use when querying postgres from a tool, modifying the binder, or adding a new DB engine.
| name | echo-sdk-models |
| description | Conversation data model — ConversationContext, Message, ToolCall/ToolResult, polymorphic content. Use when reading/writing context, adding a new content type, or debugging tool-call pairing. |
ConversationContext — root container; has messages: list[Message] and system_context.Message — has role: MessageRole and content: list (polymorphic).TextMessage, ImageContent, DocumentContent, ToolCall, ToolResult.MessageRole: user, assistant, system, tool.ContentSourceType — for image/document provenance.Provider enum in providers.py.Public API: src/echo/models/__init__.py. Implementation: user_conversation.py.
invoke() and invoke_stream() return a new ConversationContext — never mutate the input. Tools must not touch context.messages.ToolCall has a tool_call_id; the matching ToolResult must reuse the same tool_call_id. Providers rely on this.system_context.tool_context is a dict of hidden params injected into tool kwargs by the LLM layer. Not visible to the LLM. Add new hidden params here, not to tool input_schema.content: a single message can carry text + image + document + tool calls/results. Order matters for some providers (Anthropic in particular).model_validate, model_dump. For serialization to JSON, orjson.dumps(ctx.model_dump()) — never stdlib json.user_conversation.py (or a new file imported there).Message.content union (and any discriminator).src/echo/llm/<provider>.py to translate the new type to the provider's native shape — or skip it cleanly if the provider doesn't support it.tests/.context.messages from a tool → tools return values; the LLM layer appends.ConversationContext between two concurrent agent runs → make a copy or use separate contexts.Message.content → they go in system_context.tool_context (hidden) or LLM config.tool_call_id matching → provider rejects the next turn..claude/diagrams/conversation-context-er.md[[echo-sdk-tools]], [[echo-sdk-llm]]