as-goal

Orchestrates an autonomous multi-role agent pipeline that iterates until measurable exit gates pass.

Updated Apr 28, 2026
One-click install
npx skills add https://github.com/visdomtech/skills --skill as-goal-visdomtech
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: as-goal
Source: https://github.com/visdomtech/skills/tree/main/as-goal
Command: npx skills add https://github.com/visdomtech/skills --skill as-goal-visdomtech

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Complex, multi-file features often stall because vague goals, ad-hoc coding, and constant approval requests break momentum. This Skill turns a rough idea into a self-driving engineering pipeline with defined roles, immutable exit gates, and evidence-based iteration. ## Core Features & Use Cases - Goal Extraction via Interview: Runs an interview phase to produce a single measurable goal statement saved to docs/as-goals/ before any code is written. - Four-Role Agent Team: Assembles a System Architect, Senior Engineer, Test Engineer, and Delivery Lead, plus conditional bench roles (security, release, documentation, performance, frontend, debugging) activated only by trigger conditions. - Gated Iteration Loop: Iterates up to 10 times through plan, build, review, and synthesize steps, producing evidence manifests per gate and ending in DONE, POST-MORTEM, or escalation. - Use Case: You want a CSV employee import feature. The skill interviews you for the goal, defines gates like upload UI and field mapping, then loops autonomously, tracking state in PROGRESS.md until all gates pass with linked evidence. ## Quick Start Ask the AI to run the as-goal skill on your feature idea, for example: use as-goal to build a CSV import feature for employee data.

Frequently Asked Questions about as-goal

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run an autonomous multi-agent coding pipeline?▼

Invoke the as-goal skill with a feature idea. It interviews you to extract a measurable goal, assembles architect, engineer, test engineer, and delivery lead roles, defines exit gates, then iterates up to 10 times until all gates pass with linked evidence.

What is an evidence-based exit gate in agent workflows?▼

An exit gate is a measurable condition saved as a gate definition file, requiring linked artifacts like test files or implementation paths. Only the Test Engineer may mark a gate Pass, and self-reported passage without evidence is rejected.

When should I not use an autonomous goal pipeline?▼

Avoid it for simple single-file changes, exploratory work without clear success criteria, or when you need human approval at each step. Use a direct dev-cycle for small tasks and an interview skill first for unclear goals.

What happens when the iteration budget runs out?▼

After 10 iterations with failing gates, the pipeline produces a structured POST-MORTEM.md covering gate status, most frequent failure, least effective role, and a recommended scope reduction, then ends.

How does the pipeline survive restarts or context compaction?▼

A global PROGRESS.md file tracks the gate dashboard, iteration log, open defects, and next actions. On resume, the orchestrator reads PROGRESS.md first and continues from Next Actions rather than re-deriving state.