agrag.agents

Agentic layer: planner/researcher/verifier over SearchEngine.

build_agent lives in agrag.agents.build and is not re-exported here: it needs the agents extra to import, and this package must import on a base install.

Modules:

• build – Build the planner/researcher/verifier agent graph.
• errors – Errors raised by the agent layer.
• harness – Process-global DeepAgents harness profile registration.
• ledger – Citation ledger: assigns and tracks stable keys for one agent run.
• middleware – Agent middleware for composing models and bounding the research loop.
• model – Translate LLMClientConfig into the matching LangChain chat model.
• prompts – Agent prompt templates for planner, researcher, verifier, and fallback.
• result – Result type returned by an agent run.
• settings – Env-backed LLM and loop config for the agent layer.
• subagents – Subagent specs for the researcher and verifier roles.
• tools – Agent tools: thin wrappers calling SearchEngine with fixed Recipes.
• tracing – Per-run OpenInference tracing for agent runs.
• verification – The verifier subagent's structured verdict.

Classes:

• AgentLLMSettings – LLM client config for the agent's own reasoning turns.
• AgentMissingExtraError – Agent tracing needs a package extra that is not installed.
• AgentRunResult – Result of one agent run.
• AgentSettings – Configuration for the agent loop itself.
• Ledger – Assigns and tracks stable citation keys for one agent run.