scope-watch
Skill1 starsChanged 5 months agoMITView on GitHub
During execution, continuously diff the work in progress against the intent of the issue (the scope oracle), so drift is caught in flight rather than at review. Use while implementing a plan, especially across many steps.
In this repository8 other files it gives its agents.
- close-the-booksSkill
- expert-panelSkill
- gather-contextSkill
- scan-projectSkill
- seal-planSkill
- servo-flowSkill
- servo-gateSkill
- shape-specSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 25
Never widen scope silently: If the work genuinely needs more than the issue covers, that is a decision for the human (CEO/client), not a …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
scope-watch
Hold the work to its intent while it is being built, not after. A human implementer keeps the goal in their head for hours; an agent loses it as the context fills with implementation detail and starts solving adjacent problems it was never asked to. This skill catches that drift in flight.
What drift looks like
- Building something the issue did not ask for ("while I'm here, I'll also…").
- Quietly widening the change beyond the plan's scope.
- Solving a harder, more interesting problem than the one assigned.
- Touching files or systems the scope did not name.
Procedure
- Anchor on the scope oracle. From the manifest, read the
scopesource for this work (the issue / the spec's scope section). That is the intent: what this work is, and what it is not. - Diff at each meaningful step. As you implement, periodically compare the work in progress against that intent. Cheap question: does this change still serve the stated scope, or did it grow a second purpose?
- On suspected drift, stop the line. This is the andon cord: stopping is cheap and expected; pushing through a scope smell is the failure. Surface the drift — "this step is adding X, which is outside the issue's scope" — and decide explicitly: is X in scope (update the scope oracle deliberately) or out (drop it)?
- Never widen scope silently. If the work genuinely needs more than the issue covers, that is a decision for the human (CEO/client), not a default the agent takes mid-implementation.
Rule
An agent that pushes through a scope smell to "be helpful" is the exact failure this skill exists to stop. Stopping to check is the cheap, correct move.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.