Run 192 batch: three bugs, and test guidance on the real clock (#485) #486

Merged
cmoriarty merged 2 commits from run192-batch into main 2026-11-11 14:43:36 -05:00
Owner

Closes braid#485. Findings and the owner's decisions are on braid#484.

What changes

  • The text-only test rule (B11) judges each test by what it reads. In run 192, two of the three failures that made round 1 red were false.
  • The code review is offered its reader helpers. It never was: has_subagent only answered for DELEGATION_WHOLE.
  • A group working side by side keeps its saved brief in the run's tmp, which it can read.
  • The planner's sentence no longer reads as forbidding a task to run its own browser test.
  • The test guidance rules out the real clock, and says to reach a state directly, not by stepping through the whole flow.
  • The planner's and the repair's briefs say what counts toward a requirement's 500 characters; the repair gets osf.spec_length to list them, now allowed.

Checked

  • ./tools/test.sh and ./tools/sim.sh --smoke pass.
  • Brief sizes: the apply brief grows 301 characters, the new test guidance the owner asked for; the planner shrinks 4. The repair's budget rises from 1,200 to 1,400.

🤖 Generated with Claude Code

Closes braid#485. Findings and the owner's decisions are on braid#484. **What changes** - The text-only test rule (B11) judges each test by what it reads. In run 192, two of the three failures that made round 1 red were false. - The code review is offered its `reader` helpers. It never was: `has_subagent` only answered for `DELEGATION_WHOLE`. - A group working side by side keeps its saved brief in the run's `tmp`, which it can read. - The planner's sentence no longer reads as forbidding a task to run its own browser test. - The test guidance rules out the real clock, and says to reach a state directly, not by stepping through the whole flow. - The planner's and the repair's briefs say what counts toward a requirement's 500 characters; the repair gets `osf.spec_length` to list them, now allowed. **Checked** - `./tools/test.sh` and `./tools/sim.sh --smoke` pass. - Brief sizes: the apply brief grows 301 characters, the new test guidance the owner asked for; the planner shrinks 4. The repair's budget rises from 1,200 to 1,400. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
fix(lint,briefs): run 192's three bugs, and test guidance on the real clock (#485)
All checks were successful
test / suite (pull_request) Successful in 1m10s
b96a0f93ff
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
docs(openspec): tick run192-batch's tasks (#485)
All checks were successful
test / suite (pull_request) Successful in 1m10s
aedf0fc2dc
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
cmoriarty/braid!486
No description provided.