snorkel/root-repository-maintenance-s214411
2
0
Fork 0
Code Issues Pull requests 3 Projects Releases Packages Wiki Activity

deleted-file-archaeology-guide: Document Deleted File Archaeology #2

Open
std-bot wants to merge 1 commit from task/deleted-file-archaeology-guide into main
pull from: task/deleted-file-archaeology-guide
merge into: snorkel:main
Conversation 0 Commits 1 Files changed 9 +210 -97
std-bot commented 2026-06-08 21:56:10 +00:00
First-time contributor
Copy link

Assigned idea

Benchmark authoring specification.

Summary: Add a focused recipe to skills/git-archaeology/SKILL.md for finding the commit that deleted a whole file and viewing the file as it existed immediately before deletion.

Task intent: Improve the upstream git-archaeology skill so users investigating removed files have a bounded, copyable workflow that matches the existing guidance to start narrow and avoid broad git log -p output.

Benchmark plan: Construct the task package around a Markdown-only documentation improvement. The solver should update skills/git-archaeology/SKILL.md, preserve its YAML frontmatter and existing decision-tree style, and add a section for whole-file deletion archaeology with exact commands for locating deletion commits, viewing the parent version, and following prior renames.

Environment plan: The task/environment contains the instruction file and a Python verifier only. The environment/repo is the materialized pullfrog/pullfrog checkout at commit 1e27180a7064f7240e0d8b8933b795983bd6a58d; no network access or external dependencies are needed beyond Python and pytest-style execution.

Verifier plan: Create test_outputs.py that reads skills/git-archaeology/SKILL.md and asserts: the frontmatter still contains name: git-archaeology; a new decision-tree heading about deleted files exists; the file contains the exact commands git log --diff-filter=D --summary -- <path>, git show <deletion-sha>^:<path>, and git log --follow --oneline -- <path>; the new prose explains that <deletion-sha>^:<path> shows the file before deletion; and no new unscoped fenced command line uses plain git log -p without a path filter.

Oracle plan: The oracle patch should insert one Markdown subsection under ## Decision tree (by agent intent) in skills/git-archaeology/SKILL.md, using the existing heading-plus-bash-block-plus-notes format. The section should cover: identify deletion with git log --diff-filter=D --summary -- <path>, inspect the pre-deletion contents with git show <deletion-sha>^:<path>, and use git log --follow --oneline -- <path> when the file may have been renamed before deletion.

Difficulty plan: Current coding systems may over-edit the skill file, dump broad history commands contrary to the documented output discipline, omit the crucial parent-tree ^:<path> syntax, or confuse deleted-line pickaxe searches with whole-file deletion recovery.

Repo patch scope: Expected upstream edits are limited to skills/git-archaeology/SKILL.md; do not modify other skill files, repository configuration, or code directories.

Anti-leakage notes: Solver-facing materials must not mention benchmark, harness, Snorkel, Harbor, std-ts, agent, stage, attempt, retry, generated task, or evaluation context; present this only as a normal documentation improvement.

Approach: Use a minimal Markdown edit that mirrors the existing git-archaeology style: scoped command first, then a narrower inspection command, then short notes warning against broad patch logs.

Verification: Run python -m pytest test_outputs.py; the verifier checks skills/git-archaeology/SKILL.md for the deleted-file section and the concrete commands git log --diff-filter=D --summary -- <path>, git show <deletion-sha>^:<path>, and git log --follow --oneline -- <path>.

Risk: The main risk is making the verifier too wording-specific for prose. Keep exact assertions to headings and commands, while allowing explanatory text to vary as long as it states the parent-version behavior.

Path rules: task package is tasks/; mutable upstream checkout is tasks//task/environment/repo; benchmark package files are under tasks//task; verifier/container environment belongs under tasks//task/environment; regular and UI verifier logic belongs in task/tests/test_outputs.py and test.sh only runs that file through pytest.

Agent context

Task slug: deleted-file-archaeology-guide
Branch: task/deleted-file-archaeology-guide

## Assigned idea Benchmark authoring specification. Summary: Add a focused recipe to `skills/git-archaeology/SKILL.md` for finding the commit that deleted a whole file and viewing the file as it existed immediately before deletion. Task intent: Improve the upstream git-archaeology skill so users investigating removed files have a bounded, copyable workflow that matches the existing guidance to start narrow and avoid broad `git log -p` output. Benchmark plan: Construct the task package around a Markdown-only documentation improvement. The solver should update `skills/git-archaeology/SKILL.md`, preserve its YAML frontmatter and existing decision-tree style, and add a section for whole-file deletion archaeology with exact commands for locating deletion commits, viewing the parent version, and following prior renames. Environment plan: The task/environment contains the instruction file and a Python verifier only. The environment/repo is the materialized `pullfrog/pullfrog` checkout at commit `1e27180a7064f7240e0d8b8933b795983bd6a58d`; no network access or external dependencies are needed beyond Python and pytest-style execution. Verifier plan: Create `test_outputs.py` that reads `skills/git-archaeology/SKILL.md` and asserts: the frontmatter still contains `name: git-archaeology`; a new decision-tree heading about deleted files exists; the file contains the exact commands `git log --diff-filter=D --summary -- <path>`, `git show <deletion-sha>^:<path>`, and `git log --follow --oneline -- <path>`; the new prose explains that `<deletion-sha>^:<path>` shows the file before deletion; and no new unscoped fenced command line uses plain `git log -p` without a path filter. Oracle plan: The oracle patch should insert one Markdown subsection under `## Decision tree (by agent intent)` in `skills/git-archaeology/SKILL.md`, using the existing heading-plus-bash-block-plus-notes format. The section should cover: identify deletion with `git log --diff-filter=D --summary -- <path>`, inspect the pre-deletion contents with `git show <deletion-sha>^:<path>`, and use `git log --follow --oneline -- <path>` when the file may have been renamed before deletion. Difficulty plan: Current coding systems may over-edit the skill file, dump broad history commands contrary to the documented output discipline, omit the crucial parent-tree `^:<path>` syntax, or confuse deleted-line pickaxe searches with whole-file deletion recovery. Repo patch scope: Expected upstream edits are limited to `skills/git-archaeology/SKILL.md`; do not modify other skill files, repository configuration, or code directories. Anti-leakage notes: Solver-facing materials must not mention benchmark, harness, Snorkel, Harbor, std-ts, agent, stage, attempt, retry, generated task, or evaluation context; present this only as a normal documentation improvement. Approach: Use a minimal Markdown edit that mirrors the existing git-archaeology style: scoped command first, then a narrower inspection command, then short notes warning against broad patch logs. Verification: Run `python -m pytest test_outputs.py`; the verifier checks `skills/git-archaeology/SKILL.md` for the deleted-file section and the concrete commands `git log --diff-filter=D --summary -- <path>`, `git show <deletion-sha>^:<path>`, and `git log --follow --oneline -- <path>`. Risk: The main risk is making the verifier too wording-specific for prose. Keep exact assertions to headings and commands, while allowing explanatory text to vary as long as it states the parent-version behavior. Path rules: task package is tasks/<slug>; mutable upstream checkout is tasks/<slug>/task/environment/repo; benchmark package files are under tasks/<slug>/task; verifier/container environment belongs under tasks/<slug>/task/environment; regular and UI verifier logic belongs in task/tests/test_outputs.py and test.sh only runs that file through pytest. ## Agent context Task slug: deleted-file-archaeology-guide Branch: task/deleted-file-archaeology-guide <!-- std-ts:idea-pr-body -->
This pull request can be merged automatically.
You are not authorized to merge this pull request.
View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u origin task/deleted-file-archaeology-guide:task/deleted-file-archaeology-guide
git switch task/deleted-file-archaeology-guide

Merge

Merge the changes and update on Forgejo.

Warning: The "Autodetect manual merge" setting is not enabled for this repository, you will have to mark this pull request as manually merged afterwards.

git switch main
git merge --no-ff task/deleted-file-archaeology-guide
git switch task/deleted-file-archaeology-guide
git rebase main
git switch main
git merge --ff-only task/deleted-file-archaeology-guide
git switch task/deleted-file-archaeology-guide
git rebase main
git switch main
git merge --no-ff task/deleted-file-archaeology-guide
git switch main
git merge --squash task/deleted-file-archaeology-guide
git switch main
git merge --ff-only task/deleted-file-archaeology-guide
git switch main
git merge task/deleted-file-archaeology-guide
git push origin main
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
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
snorkel/root-repository-maintenance-s214411!2
No description provided.