Arnaud Héritier 2470dc87f8 fix(evaluation): read container stderr through exec.Cmd to avoid a data race ...

runDockerAgentInContainer read stderrData after cmd.Wait() while a
hand-rolled goroutine was still writing it; Wait does not wait for our
own goroutine. Let exec.Cmd own the copy via a bytes.Buffer so Wait
(bounded by WaitDelay) guarantees the data is complete before we read it.

2026-09-10 18:18:26 +02:00

..

agent_image_test.go

feat(eval): pin injected docker-agent image to host CLI version by default

2026-09-04 19:00:51 +00:00

agent_image.go

feat(eval): pin injected docker-agent image to host CLI version by default

2026-09-04 19:00:51 +00:00

assertions_test.go

feat: wire assertion runner into runSingleEval execution pipeline

2026-09-01 16:48:52 +02:00

assertions.go

feat: wire assertion runner into runSingleEval execution pipeline

2026-09-01 16:48:52 +02:00

baseline_test.go

feat: include assertions and pass@k in baseline regression comparison

2026-09-01 18:01:16 +02:00

baseline.go

feat: include assertions and pass@k in baseline regression comparison

2026-09-01 18:01:16 +02:00

build.go

feat(eval): pin injected docker-agent image to host CLI version by default

2026-09-04 19:00:51 +00:00

dockerfile_template_test.go

feat(eval): pin injected docker-agent image to host CLI version by default

2026-09-04 19:00:51 +00:00

Dockerfile.custom.template

feat(eval): pin injected docker-agent image to host CLI version by default

2026-09-04 19:00:51 +00:00

Dockerfile.template

feat(eval): pin injected docker-agent image to host CLI version by default

2026-09-04 19:00:51 +00:00

eval_test.go

feat: include assertions and pass@k in baseline regression comparison

2026-09-01 18:01:16 +02:00

eval.go

fix(evaluation): read container stderr through exec.Cmd to avoid a data race

2026-09-10 18:18:26 +02:00

judge_model_test.go

fix: restore provider registry after cf5a430d2 emptied the default

2026-06-25 13:39:08 +02:00

judge_test.go

eval: include judge reasons for passed relevance criteria

2026-04-02 14:59:39 +02:00

judge.go

feat: upgrade relevance judge prompt with rubric, CoT, and anti-bias rules

2026-09-01 09:27:26 +02:00

names.go

Add evaluation framework with Docker isolation

2026-01-08 09:22:17 +01:00

progress.go

feat: wire assertion runner into runSingleEval execution pipeline

2026-09-01 16:48:52 +02:00

save_test.go

fix(eval): always record resolved agent image in saved run JSON

2026-09-04 19:14:07 +00:00

save.go

feat(#3996): persist absolute WorkingDir provenance at every local creation path

2026-09-09 17:55:34 +02:00

scoring.go

feat: include assertions and pass@k in baseline regression comparison

2026-09-01 18:01:16 +02:00

termination.go

fix(eval): preserve budget termination outcomes

2026-07-28 15:47:36 +02:00

types.go

fix(eval): always record resolved agent image in saved run JSON

2026-09-04 19:14:07 +00:00