AI Agent News/stories/agno-releases-v3-1-6-with-eval-case-arguments-and-tool-result-search-fixes.md

Frameworks

Agno releases v3.1.6 with eval case arguments and tool result search fixes

Agno v3.1.6 lets eval cases carry run arguments and Message inputs, and changes how OS Manager answers framework questions and how offloaded tool results are searched. It also fixes copying agents or teams that use ParallelTools.

Source: github.com/agno-agi/agno/releases/tag/v3.1.6

The Agno v3.1.6 release lets eval Case objects take session_id, user_id, session_state, dependencies, metadata and knowledge_filters, which are forwarded to the agent or team run. The input field now accepts a Message, so a case can attach an image or file to the turn being graded. Cases that share a session_id run as consecutive turns.

OS Manager now queries the Agno documentation directly for framework questions instead of delegating them to Platform Manager, and delegated tasks stay scoped to the question. The access_docs tool covers both paths. Offloaded tool results gain a search_result that returns several passages from one long line, such as single-line JSON, plus a max_chars option on read_result for smaller excerpts.

A fix addresses copying an agent or team that uses ParallelTools, which AgentOS does on every request. The copy now keeps the live SDK client and receives its own settings, which removes RLock copy errors and _state finalizer warnings.

Why it matters if you build agents

Teams running evals can now set session, user, state and knowledge-filter context per case and grade image or file inputs in the same run.

Builders using ParallelTools with AgentOS should no longer see copy errors or finalizer warnings on each request.

Search over offloaded tool output now returns more than the first match, which helps when results arrive as long single-line JSON.

Get stories like this every weekday

One short email with the day's agent tooling news.

Subscribe