agentsclimarketplace

Agent brief

Skill mickzijdel/dev-hooks/plugins/dev-hooks/skills/agent-brief

Hooks and skills for Claude to write better code and verify its work, and an easy start to using Claude Code

Install
npx -y skills add mickzijdel/dev-hooks --skill agent-brief

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 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

Use when writing a task brief for an agent — dispatching a subagent, handing off work, or turning a fuzzy request into instructions — and you want it aimed at an outcome, not a method. Triggers on "brief an agent", "write a task for a subagent", "spec this out for an agent", or a handoff that says what to do but not why or what's out of scope. Pairs with the intent-check hook.

SKILL.md

3.2 KB, as published. Nobody here has run it

Agent Brief

Overview

A good brief describes the destination and how you'll know it's reached, then hands over the driving. Method-specification creeps in only when you don't trust the agent to find a good route — the fix is a sharper target and clearer fences, not more steps. Spend your words on the goal and the boundaries; leave the method box near-empty.

The skeleton

Fill these in order. METHOD is usually the smallest box — often empty.

FieldThe question it answers
GOALWhat end-state, in observable terms? (Not "improve X" — "X does Y when Z".)
WHYWhat problem does this solve / who's hurting? The agent generalizes from intent when reality deviates from the plan.
DONE WHENHow will you check it? Concrete inputs → expected results. If you can't state the check, the agent can't aim at it.
DON'TNon-goals, out-of-scope, hard constraints. The negative space is where most misfires live.
AUTONOMYProceed freely, but stop and ask if: … (calibrate to reversibility — tighter fences on auth, migrations, deletions, anything outward-facing).
PROVE ITWhat evidence back? "Ran it against the three broken inputs, pasted the output" — not "should work".
METHODUsually empty. If you must constrain the route, say why, so the reason travels to cases the rule doesn't cover.

Worked example

Weak (method-shaped, uncheckable):

Improve the error handling in the CSV parser.

Strong (outcome-shaped):

GOAL: Malformed CSV input yields a structured error naming the line and column; the parser never crashes and never swallows an error silently. WHY: Users paste broken CSVs and get a blank screen with no clue what's wrong — they should be able to self-diagnose. DONE WHEN: Feed it three broken inputs → three specific errors; existing tests still pass. DON'T: Touch the upload flow; add a dependency; change the v1 API. PROVE IT: Paste the three error outputs and the test run.

Note the strong version says nothing about how — but it's checkable, and the agent can self-correct against it before handing back.

The habit that compounds

When an agent does the wrong thing, patch the brief, not that one instance — the missing non-goal, the vague done-criterion. Reused briefs get monotonically better; hand-corrections don't. Before dispatching a big one, have the agent echo back its assumed goal + non-goals in one line (contract-first) — it catches a misread for the price of a sentence.

When NOT to use

A throwaway one-liner ("what's this function do?", "rename foo to bar") needs one line of goal, not the skeleton. Reserve the full form for handoffs where a wrong guess is expensive: subagent dispatches, migrations, anything hard to reverse.

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.