Running Tests
Run the tests and fix the failures -- test_run gives structured pass/fail the agent can act on.
What you'll see
- Launch
ryshin a project with tests; double-Esc to prompt mode. - Prompt: "run the tests and fix any failures."
- The
test_runtool executes the suite and returns structured pass/fail results; no approval to run. - Proposed fixes are approval-gated file edits; after approving, the agent reruns to confirm green.
Commands shown
rysh
run the tests and fix any failures
Keys used
EnterEscape
Transcript
0:00Run the tests and fix the failures -- test_run gives structured pass/fail the agent can act on.
0:03Start rysh inside a project that has a test suite.
0:08Double-press Escape to enter prompt mode.
0:13Ask it to run the tests and fix what fails. The test_run tool executes your suite and returns structured results -- which tests passed, which failed, and why. No approval needed to run them; the agent reads the failures directly.
0:32When the agent proposes a fix, the file edit is approval-gated -- you see the diff, approve it, and it reruns the suite to confirm everything's green.
0:42Structured test results in, fixes out -- one prompt at a time.
Key takeaway
test_run provides structured pass/fail results so the agent can fix failures and re-verify, with edits gated by approval.