The AI Toolchain — issue -024, July 26, 2026

Agno v2.8.4 adds TrustedRouter as an OpenAILike model class and revamps entity memory for the second brain.

$ git clone --branch v2.8.4 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v2.8.4

• ›Adds TrustedRouter as an OpenAILike model class, enabling use of TrustedRouter as a model backend.
• ›Revamps entity memory for the second brain, enhancing how agent memory stores and retrieves entities.

Framework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.

CrewAI 1.15.7 adds runtime skill usage event emission for observability.

$ git clone --branch 1.15.7 https://github.com/crewAIInc/crewAI.git
# already have the repo? check out this version:
$ git checkout 1.15.7

• ›Emits skill usage events at runtime for observability into agent behavior during crew execution.

$ git clone --branch 1.15.7a1 https://github.com/crewAIInc/crewAI.git
# already have the repo? check out this version:
$ git checkout 1.15.7a1

An MCP that lets AI tools securely connect to your infrastructure, write IaaS code, debug issues, and assist during incidents - without risking production stability. Built for security teams to approve and infrastructure teams to experience like magic.

emisar v0.34.0 adds streaming run output via wait_for_run cursor paging and authoring-time enforcement of pack argument substitution bounds.

$ git clone --branch v0.34.0 https://github.com/AndrewDryga/emisar.git
# already have the repo? check out this version:
$ git checkout v0.34.0

• ›Adds wait_for_run MCP call that accepts an output cursor and returns the next one, reading the event log forward; supports paging a finished run's trimmed output and fragments oversized events.
• ›Pre-run dispatch rejections (contract changes, refusals, rate limits) now log bounded, allowlisted fields, making rejected calls visible in operations without writing model input to the log.
• ›Pack-trust conflict UI now names the specific runners that disagree about an action instead of failing generically.
• ›Adds an llms.txt index and a Keycloak setup guide to the documentation site.
• ›Operator input (approval notes, grant scopes, policy overrides) now survives a console re-render triggered by a co-approver broadcast or a refused submit.