skills.rest
zts212653
zts212653COMMUNITY
Explore all skills by this author →
GitHub

open-source-teardown

Audits open-source projects by tracing marketing claims to source code evidence.

2.9k|733|Updated Mar 12, 2026
Search Tags:#architecture review#open-source#gitHub#claim verification#teardown#Competitive analysis#Code Audit
One-click install
npx skills add https://github.com/zts212653/clowder-ai --skill open-source-teardown
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: open-source-teardown
Source: https://github.com/zts212653/clowder-ai/tree/main/cat-cafe-skills/open-source-teardown
Command: npx skills add https://github.com/zts212653/clowder-ai --skill open-source-teardown

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It prevents teams from judging trending GitHub projects by README and marketing alone, replacing hype-driven evaluation with a source-code-first audit that verifies architecture, star features, algorithm substance, and feedback loops.

Core Features & Use Cases

  • Claim-to-Code Verification: Builds a claims ledger and traces every advertised feature from public API through entrypoint, state mutation, and future behavior.
  • Algorithm Peel & Feedback Loop Tests: Classifies mechanisms as true algorithms, LLM judges, heuristics, rules, or external services, and validates self-improvement claims against a signal-decision-mutation-behavior chain.
  • Structured Teardown Reports: Produces architecture maps, reproduction matrices, raw output audits, and Learn/Gap/Do-Not-Follow comparisons using templates in the references directory.
  • Use Case: When evaluating a competitor agent runtime or a viral memory/RAG framework, run this Skill to produce an evidence-backed report separating real engineering from marketing before deciding what to adopt.

Quick Start

Ask the agent to teardown a specific open-source project, for example by saying: teardown the hermes agent runtime and tell me what is real versus marketing.

Frequently Asked Questions about open-source-teardown

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

FAQPage Schema
How do I evaluate whether a trending GitHub project lives up to its marketing?▼

Use a claim-to-code audit: list each advertised feature as a verifiable claim, then trace it from public API through entrypoint, core module, and state mutation. This Skill automates that workflow and outputs an evidence-backed report.

How to verify self-improving or learning claims in an agent framework?▼

Trace the full loop: signal, decision, state mutation, and future behavior. If any link is missing, the system has telemetry or CRUD rather than a closed learning loop, and the report records that verdict explicitly.

What is the difference between this teardown Skill and general deep research?▼

Deep research gathers public sources and papers, while this Skill is a source-code-first capability audit. It clones the repository, inspects state mutations and feedback loops, and classifies algorithms rather than summarizing documentation.

Can this Skill audit benchmark or performance cost claims?▼

Yes, it builds a scoped decision ledger fixing workload, provider, model, and time window, then separates measured constructs from product decision fit. Unverifiable items are recorded as unknown rather than guessed.

When should I not use an open-source teardown?▼

Skip it when you only need literature or community research, issue and PR triage, or open-ended architecture brainstorming. Those scenarios belong to deep research, community operations, or collaborative thinking workflows instead.

Related Skills

Explore More Software Engineering Skills→
mpsuessermpsuesser

librarian

Research open-source library internals and generate GitHub permalinks for code evidence.

Community
Advanced
avgarciaavgarcia

adr-review

Review architecture decision records for missing decisions, risks, and structural gaps.

Community
Advanced
rchandnaWUSTLrchandnaWUSTL
14

Insights One-Pager

Generate a markdown one-pager from a unified Insights JSON file.

Community
Basic
mqzkimmqzkim

code-review

Review code for architecture, security, performance, and concurrency issues.

Community
Advanced