Skip to main content

Backlog

Under consideration

No Backlog issues

Next up

5

Committed and queued

Feature Request

Survive a restart

Sandboxes, runs, and the execution record persist across restarts. On startup jhansi reconciles what it has on record against what is actually running, closes out runs interrupted by the restart, and cleans up its own orphaned containers — leaving anything it did not create alone.

jhansi.io
Feature Request

Sandbox expiry and automatic cleanup

Sandboxes get a time-to-live and are reclaimed automatically once idle, so an abandoned session no longer holds a container and a working directory indefinitely. An optional absolute ceiling caps total lifetime regardless of activity. Expiry is recorded as a distinct event from user-initiated deletion. Stored run output is never removed by cleanup.

jhansi.io
Feature Request

Health and diagnostics endpoint

Expose the facts your monitoring already knows how to consume: data directory writable, free and total space, isolation engine reachable. Plus a startup check so a misconfigured data directory fails at boot with a clear message instead of at the first run. jhansi exposes signals; your existing monitoring decides the thresholds. Included in the open-source install.

jhansi.io
Feature Request

Run files in and out of a sandbox

Write files into a sandbox, read results back out, and list what a run produced. Write accepts multiple files in a single call, so an agent can push a whole set of generated files at once. Turns jhansi from "run this command" into "run this work" — write a script, execute it, fetch the output.

jhansi.io
Feature Request

MCP server

Connect any MCP client — Claude among them — directly to your jhansi install and let it create sandboxes, run code, and move files, without writing a line of HTTP. Runs inside the same single binary. No extra process, no separate service.

jhansi.io

In Progress

Actively being built

No In Progress issues

Done

2

Recently shipped