Coding tasks and agent trajectories
Each coding task comes with a reference solution and, where possible, tests. Where tests cannot judge the work, such as whether a fix addresses the real issue or whether another engineer could maintain it, an engineer grades it against a rubric.
For agent work, an engineer records each step of a task, including the reasoning, the tool call and its output.
What you can order
- Coding tasks
- Agent and tool-use trajectories
- Multi-turn coding conversations
- Rubrics for grading code
Who grows it
Working software engineers write and review every task and trajectory.
What to tell us on the order slip
- Languages, frameworks and the kind of codebase.
- Tests, a reference solution, a rubric, or some mix.
- For agents: which tools, which environment, and what finishing the task means.
- How many tasks or trajectories, and the deadline.
For agent work, the engineers record each call in your agent's own tool format. Say on the slip which tools it uses, and share the definitions once the spec is set.
How to get a free sample
- Send us an order slip.
- Agree the spec with a farmer.
- Experts grow a free pilot batch to your spec.
- Run it through your evals, then order by the crate.

