Use case

Close the loop in your software factory

Agents write code faster than anyone can check it. Composal is the verification agent in your pipeline: it tests the running feature, and sends what it finds straight back to the agent that wrote the code.

Read the docs
Coding agent writes the changePull request on a previewComposal verifies it
↺ findings go straight back to the coding agent

Where the time goes today

Fast writing, slow checking

Coding agents produce pull requests all day. Verifying each one by hand takes longer than writing it.

The author grades its own work

An agent asked to test what it just built shares its blind spots. It checks that the code does what it intended, not what users need.

No loop, only a queue

Failures end up as comments for a person to carry back to the agent, and the cycle waits on them.

1

A verifier that did not write the code

Composal sits outside the codebase and tests the product as a user, so it is not anchored on the implementation.

Writes the code
Knows how it was built
Verifies behaviour
Sees what a user sees
2

Findings go straight back

Each failure arrives with steps and evidence and is handed to your coding agent with one action, from the pull request, the CLI or an agent skill.

FailedFinding with steps and recording
Coding agent reproduces it, fixes it, and pushes a new commit.
3

Repeat until it passes

The next commit is verified the same way. People step in for the final review, not for every round.

Commit 11 finding
Commit 2Fixed, 1 new
Commit 3Passed

Verification that scales with agents

As output grows, checking keeps up without a bigger review queue.

An independent check

The thing that tests the feature is not the thing that built it.

Works from your agent’s tools

The CLI and an agent skill let your coding agent run and read verification itself.

See every use case