Tests

You are improving tests for a GitHub repository.

Environment

• The target repo is cloned at o/repo/. You have read and write access.
• You do not have access to bash. You cannot run tests — work statically.

Instructions

1. Read o/repo/AGENTS.md and any documentation it references to learn:

   • Where tests live and how they are organized.
   • What conventions exist for testing.

2. Read through the codebase and existing tests. Understand what is tested and what is not. Look for:

   • Untested code, missing edge cases, gaps in error handling.
   • Tests that exercise code but don't verify outcomes.
   • Tests that depend on external state or are otherwise fragile.
   • Any other quality issues specific to this repo's test ecosystem.

   Adapt your analysis to whatever testing patterns the repo uses. Do not impose a fixed rubric.

3. Fix the most impactful issues. Prioritize:

   • Missing tests for important code paths.
   • Weak or missing assertions.
   • Tests that are clearly broken (syntax errors, wrong imports).

   For each fix: a. Follow the repo's conventions. b. Stage and commit with a descriptive message.

Constraints

• Do not modify source files — only test files.
• Follow the repo's existing test conventions.

Output

Write o/tests/tests.json:

{
  "created": 0,
  "improved": 0,
  "findings": [
    {
      "area": "<module, feature, or test file>",
      "action": "<added|improved|flagged>",
      "notes": "<what was found and what was done>"
    }
  ]
}