01
/goal-formatter
Describe your idea in plain language; get back a structured, testable goal: measurable end state, success criteria, verification method. Self-reviews against a rubric.
Three composable AI agent skills that turn a rough idea into a hardened engineering specification, one you can trust enough to build from. Tool-agnostic, zero dependencies.
PROCESS
IDEA → HARDENED SPEC
JG SYSTEMS CONSULTING LTD.
SHEET 1 OF 1
Three skills, run in order. Full reference on the skills page; the step-by-step walkthrough is on the usage page.
01
/goal-formatterDescribe your idea in plain language; get back a structured, testable goal: measurable end state, success criteria, verification method. Self-reviews against a rubric.
02
/goal-specTurns the goal into a complete engineering spec, asking targeted clarifying questions grounded in what you said you want, resolving every gap before it's done.
03
/spec-reviewAn autonomous roundtable (Architect, Security, Test/QA, Operations, Product) critiques the spec round after round until no critical or major issues remain.
Native · SKILL.md copied as-is
Transform · converted by the installer
Pure prompt workflows: no MCP server, no runtime. Detail: docs/other-agents.md.
Plugin marketplace · Claude Code
/plugin marketplace add jgsystemsconsulting/jgs-goal-spec /plugin install jgs-goal-spec@jgs-goal-spec
Installer script · any agent
python install.py # Claude Code (default) python install.py --agent gemini # or codex / cursor / # openclaw / copilot python install.py --list-agents
The kit's deliverable is a trustworthy spec, not a running system. Once
spec-review reports convergence, hand the goal + hardened spec to an executor
of your choice: Claude Code's /goal, an autonomous loop, or an executor subagent.
The spec's Success Criteria are the acceptance gate: the work is done when every criterion is observably met. That's how the loop closes on a built outcome, not just a written one.