skill-slug

name	<skill-slug>
description	One-line description of what this fix does and why it matters.
target_harness	<harness-name>
target_repo	<owner>/<repo>
target_files	["path/to/file.ts"]
target_commit	<SHA or branch as of audit>
estimated_savings	<e.g. "30-90% input discount on Anthropic" or "0 → 80% hit rate">

Target

Which harness, which file, which lines. Cite a permalink to the relevant code at the audit commit.

Symptom

What's broken in the user's current setup. Be concrete:

• Which API call is wrong
• What field is missing/wrong/volatile
• What usage field comes back zero or wrong

Fix

--- a/path/to/file.ts
+++ b/path/to/file.ts
@@
-  old line
+  new line

If the fix is multi-step or conditional, list the steps.

Verify

After applying, the agent should:

1. Run <command> to exercise the code path.
2. Capture the wire (mitmproxy or harness debug log).
3. Confirm <specific assertion> — e.g. usage.cache_read_input_tokens > 0 on the second turn.

If the assertion fails, the fix didn't land or there's a second upstream issue. Stop and report.

Background

Why this matters in 1-2 sentences. Link to:

• [docs/concepts/.md] for the API mechanics
• [docs/gotchas.md#N] for the underlying footgun
• [audits/.md] for the full audit