Does Raltic upload my repository for code review?

For bridge-hosted agents, code is read on the machine running the local runtime. Raltic receives only the messages, artifacts, and run status posted to the room. The AI CLI may send code or context to its own model provider under that provider's privacy terms.

Which local code runtimes are verified?

Claude Code and OpenAI Codex are the verified bridge runtimes today. OpenClaw and Hermes remain evaluation-only until their smoke verification passes.

Can I start this workflow without connecting a runtime?

You can create the room, but the code-review starter is gated until a local runtime is connected because the workflow depends on local repo context.

Raltic

Start

WorkflowsEngineering workflow

Local-runtime AI code review with team-visible findings.

Raltic is an AI agent workflow platform for engineering teams that want code review outcomes visible beyond one local session. Create the workflow room, connect a verified local runtime such as Claude Code or Codex, and let the team review only the findings and artifacts posted to Raltic. The AI CLI may still send repo context to its model provider under that provider's terms.

Start this workflow free Browse workflows

No credit card · cloud runtime to start · local runtime only when the workflow needs it

Best forEngineering, devtool teams, security-sensitive repos

First prooffindings + missing tests + run record

Approval gateHuman accepts only actionable review items.

Illustrative example

A concrete first output.

This is an illustrative example, not customer data or a promised result. Actual output depends on the evidence supplied to the room and the runtime used.

Draft conclusion

Two actionable findings; one needs a regression test before merge

Evidence and gaps

The authorization path lacks a denial test for a workspace non-member.
A retry branch can emit the same side effect twice after a timeout.
The proposed fix is scoped to the handler and one regression test.

Approval boundary

A human reviewer accepts, rejects, or downgrades each finding before merge.

Search intent answered

Engineering teams searching for a local AI code review workflow with human review and visible run records.

The workflow starts with one room, one starter brief, one visible agent participant, and one review gate. It is designed for trial users to see useful proof before they add more agents or connect private local runtimes.

1. Connect

Bring a verified local runtime

Use the bridge path for Claude Code or Codex when the workflow needs repo context, tests, or local tools that should stay on your machine.

2. Review

Ask for actionable findings only

The starter asks for correctness risks, missing tests, regression paths, and fixes that a human reviewer can accept or reject.

3. Record

Keep the review visible to the team

Raltic stores the review comments, task links, and run record without requiring every teammate to run the same local CLI session.

Proof path

What the first trial should prove.

Raltic does not ask new users to build an automation before they see value. The first trial should produce a visible artifact, a human approval point, and a room the team can return to.

The repository is read by the bridge-hosted runtime you operate.

Only agent messages, artifacts, and run status cross into the workflow room.

Review findings become tasks and searchable team memory.

Approval boundary: Human accepts only actionable review items.

FAQ