(WIP) "Reasoning", "Deep ReSearch", "document_query", collapsible msgs, thought summaries #350
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
…ught summaries
The "Reasoning" button now enables agent-zero reasoning loop for current context. Possible values are on, off and auto. In on state the agent is urged to use reasoning tool, in auto mode he simply can use it if needed. The Agent can then use the reasoning_tool to perform another reasoning step. For reasoning models from OpenAI and Anthropic (native api and OR) this also enables reasoning_effort choice passing to the api.
The "Deep ReSearch" button now enables the deepsearch mode for current context. This modifies the system_prompt and explicitly instructs agent how to interview user, perform deep research etc.
The new document_query tool is able to retrieve text content via ocr from many unstructured document types including office and pictures.
The intermediate messages (thinking/generating etc.) are now collapsible. Especially with long reasoning loops or long tools outputs this helps. Left-sidebar preferences has a slider to switch from default collapsed to expanded.
"Generating" headline was changed to "Thinking", the thinking messages now have the main actoin topic in headline to make collapsed messages actually useful.
Also, squashed commit:
feat: agent-zero REASONING implementation with self-prompting for all function calling models. Opt. support for natively reasoning models. feat: improve prompts
feat: improve PROMPTS further, make them more comprehensive feat: noop_tool to allow agent to iterate on their reasoning feat: Support for an observation phase
fix: missing prompt for noop_tool
fix: improve prompts
feat: reasoning support (WIP, formerly noop_tool)
feat: Reasoning: thought summaries
Make reasoning button tristate: on off auto
feat: Autom. collapse intermediate messages slider (def. on)