agentsclimarketplace

Fog

Skill leoncuhk/defog/skills/fog

Seven tiny agent skills that burn down the gap between your prompt and reality — through one shared unknowns ledger

Install
npx -y skills add leoncuhk/defog --skill fog

Assembled 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

Maintain a FOG.md unknowns ledger for a task. Use when a task has unresolved decisions or unverified constraints, when the user mentions unknowns or the fog ledger, or when another fog-* skill needs the ledger rules.

SKILL.md

3.0 KB, as published. Nobody here has run it

Fog

Fog is the gap between the map (prompt, plan, assumptions) and the territory (codebase, APIs, real constraints, user taste). The ledger makes fog visible and burns it down to a defined end.

The ledger

One file per task: FOG.md, next to where the work happens (or where the user says). When a second task joins the directory, rename the existing ledger to FOG-<task-slug>.md before creating the new one, also slugged.

# FOG: <task>

## Open
- [KU] <fact to verify> → lookup
- [KU] <decision the user must make> → ask
- [UK] <taste the user will recognize on sight> → show
- [UU?] <suspected constraint nobody has checked> → scout

## Assumed
- [A] <default taken> — cost to revisit: <cost>

## Resolved
- [x] <decision or finding> (<who: user | territory>, <date>) — <why>

## Deviations            <!-- add on first deviation; never keep it empty -->
- <plan said / did instead / why / cost to revisit>

territory covers anything verified against the real system, including your own experiments. Each arrow names the resolver: → lookup you do yourself; the rest are the matching fog-* skill.

Quadrants and their resolvers

TagFog typeResolver
Known knownNot fog. Straight into the spec.
KUKnown unknown, fact→ lookup — the territory answers it. Never ask what code or docs can answer.
KUKnown unknown, decision→ ask (fog-ask)
UKUnknown known — taste, recognized only on sight→ show (fog-show)
UU?Suspected unknown unknown→ scout (fog-scout)
ALow-risk residueTake the most reversible default and label it.

Rules

  • Every open item carries its resolver. The ledger is a work queue, not a diary.
  • Burn-down is visible. Announce <n> open, <m> assumed after every ledger write. Planning is complete when every open item is resolved or converted to a labeled assumption.
  • Assumptions are vetoable, not silent. Present them as one batch; never interleave with blocking questions.
  • Resolved keeps its why — decisions and territory findings that shaped the plan. A finding shaped the plan only if a later step would act differently knowing it; the rest is dropped.

Lifecycle

  • Birth: any fog-* skill that would write to a missing ledger (or section) creates it first.
  • Handoff: the ledger plus the persisted plan (see fog-plan) is the launch packet — hand both to any fresh context or subagent.
  • Death: when a fog-quiz round passes, or the user declares the task done, graduate the whys that are hard to reverse or surprising without context into ADRs, docs, or the commit message, then delete the ledger. It is working memory, not documentation.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.