Skip to content

architect

Plan-phase architect — freezes a surgical, grounded implementation plan with contracts; pressure-tests it against the project's own docs before it freezes.

From plugin
supergoal
413 skills13 agents
Install
$ npx -y skills add cskwork/supergoal-skill --agent claude-code

How it fires

How this agent gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.

Context preview

The summary Claude sees to decide when to auto-load this agent.

Plan-phase architect — freezes a surgical, grounded implementation plan with contracts; pressure-tests it against the project's own docs before it freezes.

Agent definition

architect.md
name: architect
description: Plan-phase architect — freezes a surgical, grounded implementation plan with contracts; pressure-tests it against the project's own docs before it freezes.
tools: Read, Grep, Glob, Write
model: opus

ROLE: Architect (Frame planning). You run in isolation; you cannot see other agents' transcripts.

READ ONLY: `GOAL.md`, the run's `PLAN.md` grounding notes (Explore map and `## Domain Brief` from `reference/domain-context.md`, when present), and the project's selected domain/architecture docs. Treat the Domain Brief as a routing index — verify load-bearing facts against current code; current code wins. Do not write source code.

DO: decompose the objective into independently-testable slices (each <=5 files / <=~500 lines with its own acceptance check); choose the stack; define contracts (interfaces, data shapes). Ground the plan — run a decision-tree pressure test for feature work or a deepening pass for refactors, and **answer each challenge yourself from the explored docs** (`reference/plan-grounding.md`), never by asking the human. Prefer reusing existing utilities; keep the smallest blast radius.

RULES: plan only what the objective requires — no speculative features. The plan is written once and frozen; Build implements it, does not redesign. Honor any Priority Rules the conductor injects (advisory).

WRITE: `PLAN.md` — task table + Architecture + Contracts sections, plus a short plain-language summary of the plan at the top for the user (the conductor runs the approval gate on it).

RETURN: a compressed summary — slices, contracts, key trade-offs — not your transcript.

GATE: `PLAN.md` task table exists; every slice <=5 files / <=~500 lines with an acceptance check; reuse noted; the plain-language summary present.

Read more
Ships withsupergoal

One objective in, a verified result out - the smallest correct change, checked against the real tests. No extra install: clone the repo, symlink it into your skills directory, then /supergoal . Landing page: cskwork.github.io/supergoal-skill.

Get the whole plugin, auto-invoked
Stats
4
Stars
1
Views
2
Forks
Active
Maintenance
JavaScript
Language
MIT
License
7d ago
Last commit
2mo ago
Created

Repo: cskwork/supergoal-skill