Raider / Agent workflows for engineering
Turn repository investigation, code changes, checks, and reviews into repeatable workflows. Raider brings coding agents into your engineering process through Scala, a command-line runner, and CI.
Developer preview · MIT license · Runs from a source checkout.
Run a baseline check against a small billing fixture.
Ask an agent to change the code, then run an independent compiler and test check.
Collect source and boundary reviews, a structured decision, and run artifacts.
Where Raider can help
Ask an agent to inspect code, locate relevant files, and explain how a project works before making a change.
Give a coding agent a bounded task and a validation command. Inspect the resulting diff and actual check output.
Compose parallel agent requests for different review questions and combine their results in a typed workflow.
Connect repair, compiler and test checks, and review stages. Produce a structured outcome and traceable run artifacts.
The published CI example demonstrates one small fixture. Suitability and performance on your repositories need workload-specific validation.
One part of a broader agent ecosystem
Connect through supported OpenAI-compatible or Anthropic-compatible transports. Work from the Scala REPL or headless CLI, using the same runtime for composed workflows.
Getting started →Route model requests through the gateway to apply configured access and budget policies and inspect usage. Other supported agents can use the same model infrastructure.
Explore YABANIN →Know what produced the result
Review summaries, structured results, event logs, and transcripts. Lineage artifacts describe the recorded call graph and connect steps to their details.
Read a run’s lineage →Your checks and review process determine whether work is accepted. Raider records observed and uncertain model costs and identifies unpriced attempts when accounting is available.
Agents execute with the permissions you configure; repository isolation and release approval remain part of your engineering environment.
Put the platform to work
Bring a recurring engineering task. We’ll discuss the workflow, the checks it needs, and how to measure its usefulness.