Fog scout
Seven tiny agent skills that burn down the gap between your prompt and reality — through one shared unknowns ledger
npx -y skills add leoncuhk/defog --skill fog-scoutAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 13 days oldThe repository was created 13 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Sweep unfamiliar territory for unknown unknowns before the plan hardens. Use when the user enters an unfamiliar codebase area or domain, asks for a blindspot pass, or a FOG.md item is tagged → scout.
SKILL.md
1.5 KB, as published. Nobody here has run it
Fog Scout
Go look for what the user doesn't know exists: the module and its history, conventions and half-migrated patterns, documented limits of load-bearing dependencies, prior art in and out of the repo. The territory is read-only — record findings, never repair them.
Deliver the report in your reply, four sections, items ranked within each by how much they would change the plan:
- Landmines — mistakes a newcomer here typically makes.
- Hidden constraints — decisions already made that bound the work; invariants that must hold.
- What good looks like — 2–3 concrete examples to calibrate against.
- Questions an expert would ask — with your best-guess answer for each.
Then write the findings into the ledger (fog skill): territory facts to Resolved with their why, defaults to Assumed, remaining questions to Open with resolvers — including new UU?s for a deeper scout. Confirmed-safe areas are dropped unless a later step would act differently knowing them.
End by rewriting the user's original request as the territory shows it should read, and make it the ledger's # FOG: title — the map, corrected.
Scouting ends at understanding — no implementation. "No significant fog here" is a valid, valuable report.