Project-Xray
KalWu33/ProjectXray
Three Hermes skills that help non-programmers understand, run and modify unfamiliar software projects
Project-Xray is a set of three Hermes skills that analyze an unfamiliar GitHub repository, website or piece of software and build a persistent knowledge base about it. It is aimed at vibe coders who rely on AI agents and want to understand how a project works.
What Project-Xray does
The system has a router skill, project-mentor, which tracks what is known and decides the next step; project-quick-scan, which judges whether a project is worth deeper study; and project-deep-analysis, which does the investigation, reverse engineering and modeling. Inputs can be a GitHub repository, a website or an application. Findings are sorted into facts, hypotheses, contradictions, unknowns, open questions and decisions, and each claim is tied to evidence with a strength rating and a confidence level.
Analysis depth runs across six levels, from orientation through usage, runtime, implementation and engineering to agent context, and the user's intent, such as use, deploy, modify or fork, selects which levels apply. Output comes in three layers: a guided explanation in chat, a PROJECT_KNOWLEDGE.md file for people, and an AGENTS.md or PROJECT_CONTEXT.md file that helps a coding agent change the project safely. State is persisted in STATE.yaml and EVIDENCE.yaml and updated incrementally. The README is written in Chinese.
Key features
- Three skills: project-mentor, project-quick-scan and project-deep-analysis
- Knowledge state split into facts, hypotheses, contradictions, unknowns, open questions and decisions
- Evidence strength and claim confidence recorded for each conclusion
- Six analysis depth levels chosen by user intent
- Persistent STATE.yaml and EVIDENCE.yaml files updated incrementally
- Outputs a human knowledge file and an AGENTS.md for coding agents
When to use it
- Deciding quickly whether an unfamiliar GitHub project is worth studying
- Working out how to deploy or modify a project you did not write
- Producing an AGENTS.md so a coding agent can edit the project safely
Who it is for: Vibe coders with limited programming background who use Hermes Agent to understand and adapt other people's projects.
How it fits with Hermes Agent
All three components are Hermes skills, tagged hermes-skill, and the README frames the output as context for coding agents such as Claude Code, Codex and Cursor.
FAQ
What is Project-Xray?
Project-Xray is a reverse-understanding and tutoring system for unfamiliar projects, delivered as three Hermes skills. It investigates a repository, website or application and builds up a persistent knowledge base about it.
Does Project-Xray work with Hermes Agent?
Yes, it is packaged as three Hermes skills: project-mentor, project-quick-scan and project-deep-analysis. The mentor skill routes requests and tracks what is known, while the deep-analysis skill does the investigation.
Is Project-Xray free and open source?
Yes. The repository is MIT licensed.
Similar skills for Hermes Agent
All skillsAgent skill that guides you through writing academic papers and theses, section by section
k1m0206 ChatGPT Image Gen SkillHermes skill that generates images through the ChatGPT web UI in a logged-in Chromium browser
julien-ducerf Craft Image PromptsPortable Agent Skill that turns a visual brief into a precise prompt for image generation or editing
Visol-456 gree-ac-controlSkill and Python CLI for controlling Gree WiFi air conditioners over the local network, with no cloud
Rishiidev Hermes GitHub LaunchHermes skill that launches a GitHub repo with parallel subagents, social preview, Pages and a release
lazyperfectionist Hevy Hermes Agent SkillHermes skill and dependency-free Python CLI for Hevy workout analysis, routine management and webhooks
Related guides: What is Hermes Agent? · SOUL.md for Hermes Agent: what it is and how to write one