agent-dx

After uploading Opus adoption reports run locally through , the site showed two Opus matrices: one for and one for . Group Adoption by the model name with the provider prefix dropped, the way Practical already does (#42), so the newest report per cell wins regardless of the provider it ran through. [ ] Add tests [x] Run tests [x] to format the code [ ] Add TSDoc/JSDoc to document the code

In the 09-07 weekly matrix, 51 of 80 Opus adoption runs failed with from the AI Gateway even at concurrency 1, so the site's Opus adoption cells show "Failed". A 429 before the agent has touched the workspace is not a measurement; the runner now waits (15s, 45s, 90s) and starts the conversation again, up to three times, and records in the run metrics (optional field, no schema bump). A rate limit mid-run still fails the run, since replaying tool calls on a modified workspace would not be the same measurement. [ ] Add tests [x] Run tests [x] to format the code [x] Add TSDoc/JSDoc to document the code

On the site, was a bare id with success and tokens, so the thing it measures — baseline agents starting and hanging for 600s while cli+skill finishes in 30s — was invisible. Practical reports now carry (optional field, no schema bump), the site prints it under the task id, and every cell adds the median duration. Also sharpened the Workers task description. [ ] Add tests [x] Run tests [x] to format the code [x] Add TSDoc/JSDoc to document the code