Skip to content

RLM Code v0.1.12: Experimental Native Coding Agent

Choose a tag to compare

View all tags
@Shashikant86 Shashikant86 released this 09 Aug 19:43
· 2 commits to main since this release
v0.1.12
d71f8a0

RLM Code v0.1.12

This release introduces an experimental native coding-agent workflow for working directly on a repository
from the command line.

Give the agent a coding task and it can inspect the project, update files, run tests, and show its progress
as it works.

Highlights

  • Start repository coding tasks with rlm-code agent run
  • Maintain Python variables and working context between model turns
  • Search, read, and update files under the configured approval policy
  • Run project commands and tests through the selected sandbox
  • See live Python activity, actions, approvals, usage, and final results
  • Cancel work or constrain it with turn and time limits
  • Resume saved root-agent sessions and supported Python state
  • Continue using the existing research, benchmark, replay, and TUI workflows

Quick start

rlm-code agent run "fix the failing tests and add a regression test" \
  --repository . \
  --sandbox docker \
  --approval confirm-high

Pass --model provider/model to choose a model explicitly, or omit it to use your configured default.

Resume a session using the ID displayed by the previous run:

rlm-code agent run "continue the task and rerun verification"
--repository .
--sandbox docker
--resume

Sandbox guidance

Docker is the recommended execution sandbox. Apple Container is also suitable when configured.

The local runtime is intended for development in trusted environments and does not provide a container
isolation boundary. Use --approval confirm-high or --approval confirm-all when working with repositories
you do not fully trust.

Current limitations

The native coding agent is experimental and currently operates as a single root agent. This release does
not yet include:

  • Concurrent child agents
  • Parent/child communication
  • A live agent-tree view
  • Replay or resume of a complete multi-agent hierarchy

Upgrade

Using uv:

uv tool upgrade rlm-code

Using pip:

pip install --upgrade rlm-code==0.1.12

Full changelog: https://github.com/SuperagenticAI/rlm-code/blob/main/CHANGELOG.md
PyPI: https://pypi.org/project/rlm-code/0.1.12/

Assets 2
Loading