What does the Accept No Substitutes skill do?

It monitors Claude's output for markers of incompleteness—like 'TODO', 'temporary', or 'mock-data'—and forces the AI to ask the user for specific details instead of generating incomplete code.

Does this skill scan my existing codebase?

No, this skill is specifically designed to validate what the AI agent produces in real-time. It ensures that any new or modified code being written is complete, but it is not a general-purpose linter for your existing files.

What happens when a forbidden token is detected?

The skill immediately halts the generation process, discards the substituted content, and triggers an interview template that asks the user for the actual implementation details or missing logic.

Which tokens are considered forbidden?

It targets four main categories: Prefix substitutions (mock-*, fake-*), completeness evasions (TODO, FIXME, placeholder), deferral signals (later, eventually, for now), and example/demo markers.

Accept No Substitutes

Name: Accept No Substitutes
Author: plurigrid

byplurigrid

0•

生产力与工作流

Eliminates placeholder tokens and incomplete implementations from AI output by enforcing real-world code generation.

Accept No Substitutes is a high-integrity validator skill for Claude Code designed to prevent the accumulation of technical debt caused by lazy AI outputs. It acts as an automated quality gate that scans agent output for 'placeholder' patterns—such as TODOs, mock data, and pseudo-code—and immediately halts the process if incompleteness is detected. Instead of accepting deferred decisions, it triggers a mandatory user interview to gather missing requirements, ensuring that every line of code generated is production-ready, fully realized, and free of temporary stubs.

主要功能

010 GitHub stars

02Real-time detection of placeholder tokens like TODO, mock-*, and pseudo-*

03Mandatory user interview trigger to resolve ambiguity during code generation

04Multi-phase validation pipeline including Regex, AST structural checks, and compression analysis

05Zero-tolerance enforcement for 'technical debt laundering' and deferred logic

06GF(3) integration for validator/generator/coordinator mediation

使用场景

01Ensuring AI-generated pull requests are complete and free of stubs or mock services

02Preventing the use of 'example_' or 'demo_' configurations in production-critical files

03Forcing explicit requirement gathering when an agent encounters architectural uncertainty

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi accept-no-substitutes

For use in Claude.ai and ChatGPT

Download Skill