project-understanding

Scan a repository and generate evidence-grounded JSON and Markdown project documentation.

Updated Aug 1, 2026
One-click install
npx skills add https://github.com/QuyDu/Skills-Orchestrator --skill project-understanding-quydu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: project-understanding
Source: https://github.com/QuyDu/Skills-Orchestrator/tree/main/.github/skills/project-understanding
Command: npx skills add https://github.com/QuyDu/Skills-Orchestrator --skill project-understanding-quydu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Keeping project documentation accurate is hard: READMEs drift from code, architecture descriptions go stale, and onboarding guides contradict actual behavior. This Skill performs a complete repository rescan and rebuilds an authoritative, evidence-grounded project guide so documentation always reflects the current code. ## Core Features & Use Cases - Full Repository Scan: Walks the entire repository with bounded exclusions, hashes every scanned file, and records a repository digest for provenance. - Evidence-Grounded Reporting: Inventories skills, prompts, agents, schemas, commands, features, and workflows, each backed by verified repository-relative evidence paths. - Atomic Dual Output: Publishes reports/project-understanding.json and reports/project-understanding.md as one atomic pair, preserving the last complete version if publication fails. - Use Case: After shipping a feature, run the scan to rebuild the contributor guide, then let downstream skills like project-video use the rebuilt JSON as the authoritative narration source. ## Quick Start Ask the agent to run the project-understanding scan from the repository root to rebuild the JSON and Markdown project guide.

Frequently Asked Questions about project-understanding

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

FAQPage Schema
How do I generate project documentation from my repository automatically?

Run the project-understanding scan script from the repository root with Node.js. It walks the codebase, inventories skills, prompts, agents, and commands, then writes a validated JSON report and a derived Markdown guide to the reports directory.

How do I keep a project guide in sync with current code?

Treat each run as a full rebuild rather than an incremental update. The scanner hashes every file, records a repository digest, and atomically replaces both output files so the guide never mixes stale and new content.

Does the repository scan read secrets or sensitive files?

No. The scanner excludes .env files, private keys, certificates, and credential manifests, skips symbolic links and dependency trees, and never copies secret values or binary content into the generated reports.

What happens if the repository has too little content to document?

The scan returns a blocked status when the repository is an empty baseline or lacks enough evidence to explain its purpose and use. It records limitations instead of inventing behavior, and validation fails closed on missing evidence.

What files does the project understanding scan produce?

It produces reports/project-understanding.json validated against a schema and reports/project-understanding.md generated from the same record. Both are published atomically as one pair, with rollback to the previous version if publication fails.