Automate end-to-end feature development with a verifiable goal-loop.
npx skills add https://github.com/kenlck/skills --skill feature-dev-auto
Please help me install this Agent Skill. Skill: feature-dev-auto Source: https://github.com/kenlck/skills/tree/main/ken-swe/skills/feature-dev-auto Command: npx skills add https://github.com/kenlck/skills --skill feature-dev-auto
Front-loads decision-making for substantial feature work and enforces a verifiable goal loop, reducing drift, miscommunication, and rework by guiding implementation through staged architecture, tests, and evaluative checks.
Provide a plan for a new feature using the feature-dev-auto workflow and specify the desired iterations.
High-intent search queries and answers about installing and using this skill.
Autonomous feature development is automated using a verifiable goal-loop that applies runnable checks, explicit assertions, and must-not-change constraints to ensure disciplined progress. This enforces automated validation and iterative refinement throughout implementation.
Front-loading decisions for uncertain features requires interactive stages like Discovery, Exploration, and Architecture to establish clear goals and constraints. This staged approach reduces drift, miscommunication, and rework before implementation begins.
A verifiable goal loop is implemented by satisfying runnable checks, concrete assertions, and must-not-change constraints during Stage 3 planning. This ensures automated validation evaluates surgical changes and self-corrects until checks pass green.
Autonomous feature development suits substantial, uncertain features that benefit from automated validation and iterative refinement. It guides implementation through an automated loop of surgical changes, evaluation, and self-correction until all checks pass.
Unresolved issues during automated implementation are handled through built-in escalation paths. The automated implementation loop performs surgical changes, evaluates results, and attempts self-correction until green, escalating only when issues remain unresolved.
A verifiable goal loop is needed for substantial, uncertain features requiring front-loaded decisions and automated validation. It reduces drift and rework by enforcing runnable checks and must-not-change constraints, which manual planning lacks.
saffron-healthGenerate phased specification documents from code and documentation analysis.
nicholls-incConvert verified Dafny code into Python or Go.
nanorepublicaStandardize unit-test coding practices across Claude Code projects.
mtsatryanAutomate goal-directed planning, adaptive learning, and self-correction for complex tasks.
