Jev in production › Agent tool and action selection

Optimus

Rust coding harness with opt-in Jev feature gates that withdraw or filter tools, adjust reasoning effort, filter context, memory and code-search results, and judge loop control, retries and verification; the host validates every decision.

Open on GitHub ↗

The source publishes no measured number.
Use
Agent tool and action selection
Industry
Developer tools
Form
Open-source tool
Stage
In production
Listed
2026-09-24
Found via
discord
Repository
telemusai/optimus-agent
Stars
14
Forks
1
Last push
2026-10-10
Language
Rust
License
MIT
Optimus screenshot
docs/images/optimus-agent.png in the telemusai/optimus-agent README, MIT; shown from GitHub.

The README opens with

A self-improving, multi-model harness for coding, research, and long-running work, implemented in Rust with a persistent Python runtime.

Optimus is the layer between an AI model and real work: the tools it can use, the context it retains, the agents it coordinates, and the state it recovers when a session ends or a connection drops.

It combines a persistent Python workspace, recursive subagents, inspectable memory, provider-aware context management, and background execution. Use one model to plan and review, others to investigate or build, and keep the work connected across sessions.

Badge

For the project's own README, linking back here:

Listed in Jev in production

Also used for agent tool and action selection