snorkel/root-repository-maintenance-8ac8ec1620d541
2
0
Fork 0
Code Issues Pull requests 5 Projects Releases Packages Wiki Activity

fallback-checklist-and-safety: Add preflight checklist and safety rules to SKILL.md #1

Open
std-bot wants to merge 1 commit from task/fallback-checklist-and-safety into main
pull from: task/fallback-checklist-and-safety
merge into: snorkel:main
Conversation 0 Commits 1 Files changed 1 +36
std-bot commented 2026-06-07 00:02:32 +00:00
First-time contributor
Copy link

Assigned idea

Strengthen the AI-facing manual by adding a clear, scannable preflight checklist (API/CLI/file-edit/browser-automation first), explicit “when not to use” guidance, and concrete safe/unsafe GUI examples so agents only invoke clawdcursor as a true last-mile fallback.

Approach: 1) Edit SKILL.md to add a new top-level section near the existing “FALLBACK ONLY” guidance that includes:

  • A short preflight checklist that must be satisfied before invoking the skill (native API, CLI, direct file editing, existing browser automation).
  • A “When NOT to use this skill” subsection with concrete examples (e.g., if the goal can be achieved via API/CLI; if the user hasn’t explicitly requested GUI driving).
  • A “Safe usage rules” subsection that clarifies: require explicit user intent for GUI control; minimize on-screen exposure; prefer copy/paste of only necessary text; stop on unexpected dialogs/permissions prompts.
  1. Add an “Examples” subsection mapping common user phrases (e.g., “open X”, “click Send”) to acceptable invocations, and contrasting examples where the agent should choose API/CLI/file edits instead.
  2. Ensure the new sections are concise, numbered, and easy for an LLM to follow (consistent formatting, clear MUST/SHOULD language).

Verification: Confirm SKILL.md contains the new sections and example blocks by running: grep -n "Preflight checklist" -n SKILL.md and grep -n "When NOT to use" -n SKILL.md, then visually review the updated sections with sed -n '1,200p' SKILL.md to ensure the checklist and examples are present and unambiguous.

Risk: Low risk (documentation-only). Primary risk is over-restricting usage guidance; mitigate by keeping examples balanced and aligning language with the existing “FALLBACK ONLY” intent.

Agent context

Task slug: fallback-checklist-and-safety
Branch: task/fallback-checklist-and-safety

## Assigned idea Strengthen the AI-facing manual by adding a clear, scannable preflight checklist (API/CLI/file-edit/browser-automation first), explicit “when not to use” guidance, and concrete safe/unsafe GUI examples so agents only invoke clawdcursor as a true last-mile fallback. Approach: 1) Edit SKILL.md to add a new top-level section near the existing “FALLBACK ONLY” guidance that includes: - A short preflight checklist that must be satisfied before invoking the skill (native API, CLI, direct file editing, existing browser automation). - A “When NOT to use this skill” subsection with concrete examples (e.g., if the goal can be achieved via API/CLI; if the user hasn’t explicitly requested GUI driving). - A “Safe usage rules” subsection that clarifies: require explicit user intent for GUI control; minimize on-screen exposure; prefer copy/paste of only necessary text; stop on unexpected dialogs/permissions prompts. 2) Add an “Examples” subsection mapping common user phrases (e.g., “open X”, “click Send”) to acceptable invocations, and contrasting examples where the agent should choose API/CLI/file edits instead. 3) Ensure the new sections are concise, numbered, and easy for an LLM to follow (consistent formatting, clear MUST/SHOULD language). Verification: Confirm SKILL.md contains the new sections and example blocks by running: `grep -n "Preflight checklist" -n SKILL.md` and `grep -n "When NOT to use" -n SKILL.md`, then visually review the updated sections with `sed -n '1,200p' SKILL.md` to ensure the checklist and examples are present and unambiguous. Risk: Low risk (documentation-only). Primary risk is over-restricting usage guidance; mitigate by keeping examples balanced and aligning language with the existing “FALLBACK ONLY” intent. ## Agent context Task slug: fallback-checklist-and-safety Branch: task/fallback-checklist-and-safety <!-- 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/fallback-checklist-and-safety:task/fallback-checklist-and-safety
git switch task/fallback-checklist-and-safety

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/fallback-checklist-and-safety
git switch task/fallback-checklist-and-safety
git rebase main
git switch main
git merge --ff-only task/fallback-checklist-and-safety
git switch task/fallback-checklist-and-safety
git rebase main
git switch main
git merge --no-ff task/fallback-checklist-and-safety
git switch main
git merge --squash task/fallback-checklist-and-safety
git switch main
git merge --ff-only task/fallback-checklist-and-safety
git switch main
git merge task/fallback-checklist-and-safety
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-8ac8ec1620d541!1
No description provided.