- Generated by
1.9.8
|
NeoGraph 0.16.0
A C++20 Graph Agent Engine Library — LangGraph for C++
|
Configuration knobs for the Deep Research graph. More...
#include <deep_research_graph.h>
Public Attributes | |
| std::uint64_t | brief_max_output_tokens = 800 |
| The output cap of each step. | |
| ProviderControls | controls {} |
| Controls every step sends with its model call (temperature, reasoning effort, routing, ...), beside its output cap. | |
| bool | enable_clarification = false |
When true, inserts a clarify node between __start__ and brief. | |
| bool | enable_human_review = false |
When true, inserts a human_review node between final_report and __end__. | |
| int | max_concurrent_researchers = 2 |
| Cap per conduct_research fan-out batch. | |
| int | max_researcher_iterations = 2 |
| Inner LLM↔tools loop cap per researcher. | |
| int | max_supervisor_iterations = 2 |
| Supervisor planning rounds. | |
| std::string | model |
| The model every step calls. Required. | |
| RetryPolicy | retry {2, 15'000, 2.0f, 45'000} |
| The node retry policy of every step that calls the model: by default twice, after 15 s and 30 s. | |
Configuration knobs for the Deep Research graph.
The graph names no model and sends no sampling of its own: model is required, and controls holds what every step sends (only the members that are set). Leave controls.temperature unset for a model that refuses it (gpt-5*, o3, Claude 4.7 and later). The iteration and concurrency defaults are small, to stay inside low rate-limit tiers; raise them where the provider allows more.
Definition at line 33 of file deep_research_graph.h.
| std::uint64_t neograph::graph::DeepResearchConfig::brief_max_output_tokens = 800 |
The output cap of each step.
A supervisor, researcher, compression or report step that stops at its cap having written nothing asks again with twice the cap, up to max_output_budget.
Definition at line 44 of file deep_research_graph.h.
| ProviderControls neograph::graph::DeepResearchConfig::controls {} |
Controls every step sends with its model call (temperature, reasoning effort, routing, ...), beside its output cap.
Unset members are not sent. controls.max_output_tokens is ignored: the caps below apply.
Definition at line 41 of file deep_research_graph.h.
| bool neograph::graph::DeepResearchConfig::enable_clarification = false |
When true, inserts a clarify node between __start__ and brief.
The node asks the LLM whether the user's query is specific enough to research; if not it throws NodeInterrupt with a clarifying question, expecting the caller to resume() with the user's answer. On resume the answer is appended to the user_query and the run proceeds to the brief node. This is the analogue of langchain-ai/open_deep_research's clarify_with_user step.
Definition at line 77 of file deep_research_graph.h.
| bool neograph::graph::DeepResearchConfig::enable_human_review = false |
When true, inserts a human_review node between final_report and __end__.
The node throws NodeInterrupt on first execution so the engine saves a checkpoint and the caller can show the report to a human; on resume it inspects the latest user message in the messages channel and either:
__end__ if the message is empty / "approve" / "ok"supervisor_messages and routes back to supervisor (with supervisor_iterations reset) so the agent can address the feedback in another research round. Pairs naturally with PostgresCheckpointStore for cross-process resume — see examples/26_postgres_react_hitl. Definition at line 67 of file deep_research_graph.h.
| RetryPolicy neograph::graph::DeepResearchConfig::retry {2, 15'000, 2.0f, 45'000} |
The node retry policy of every step that calls the model: by default twice, after 15 s and 30 s.
It retries an ordinary failure of the step. The provider runtime retries a transient provider failure (a rate limit, an overload) itself, and a refused request (a wrong model id, a refused parameter) fails the run.
Definition at line 54 of file deep_research_graph.h.
1.9.8