Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →tobihagemann/recall-reasoning
Recall the reasoning behind a past change by locating the Claude Code transcript that produced it.
Passed 11 of 17 test cases with the skill, and 12 of 17 without it.
2 fixed13 passed either way5 still failing3 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 2 test cases the skill fixed. Same model, same prompt, with and without the skill.
For commit `3a4b5c6` in `lib/parser.rs`, we retrieved session reasoning from `~/.codex/sessions`. What is the recommended length limit for the narrative explanation?
| Checked | Without skill | With skill |
|---|---|---|
| Limits the explanation to two paragraphs or fewer | Missed | Passed |
The instructions your model reads when the skill loads.
Recover the reasoning behind a change. Prefer Codex session history when it can be found; otherwise derive the explanation from git history and current code.
Accept any of:
<path>:<line>)If only a file is given, use git blame to resolve the commit that last touched the line.
Resolve the target commit:
git rev-parse <sha>
git blame -L <line>,<line> -- <path>
Collect the commit subject, changed files, and relevant diff:
git show --stat --oneline <sha>
git show -- <path>
If Codex session files are available, search them for the commit SHA, touched file paths, branch name, and distinctive user request text:
Write the patterns to a file with apply_patch, one per line, then match them literally:
rg -F -f <pattern-file> ~/.codex/sessions
test cases passed, vs 12 / 17 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.