Skip to content

Reorder benchmark builder guidance #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
skrsteski merged 1 commit into from
Sep 27, 2026
Merged

Reorder benchmark builder guidance #3

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion docs/builders.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ Each builder translates one pinned benchmark into task, verifier, and result row
| `Record` | One verifier result, trial result, or source-reported summary |
| `Trajectory` | Step-by-step interaction between the agent and the environment during the task |

Below you will find an example for adding a benchmark. All data models and important fields can be found in [`models.py`](../src/messier/models.py), while more information is available in the existing builders' READMEs under [`builders/`](../src/messier/builders). Typically, there are several things to check:
Typically, there are several things to check when adding a new source:

1. **Environment**
- What state type is it? Choose one of `none` (no state beyond the task inputs), `filesystem` (files and directories, such as a software repository), `in_memory` (program objects, such as a tool-managed database), or `live_web` (an external website or web service). Only one environment state type is stored per task for now.
Expand Down Expand Up @@ -42,6 +42,7 @@ Below you will find an example for adding a benchmark. All data models and impor
- Did an intermediary collection, such as BRIDGE or Agent Psychometrics, provide the data? If so, store it in `data_provider`.
- Is every repository commit or dataset revision pinned so the source can be fetched again?

Below you will find an example for adding a benchmark. All data models and important fields can be found in [`models.py`](../src/messier/models.py), while more information is available in the existing builders' READMEs under [`builders/`](../src/messier/builders).

```python
from ...models import (
Expand Down