Verified:GitHub

deep-code-enrichment

Generates one-line LLM descriptions for unenriched code symbols in a code index.

Updated Apr 20, 2026

Search Tags:#documentation #semantic search #code index #hero #code scanning #code enrichment #symbol descriptions

One-click install

npx skills add https://github.com/astroville/sprout --skill deep-code-enrichment-astroville

Or copy as Structured Prompt for Agent▼

Please help me install this Agent Skill.
Skill: deep-code-enrichment
Source: https://github.com/astroville/sprout/tree/main/.claude/skills/deep-code-enrichment
Command: npx skills add https://github.com/astroville/sprout --skill deep-code-enrichment-astroville

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Code indexes built by static scanning contain symbols without human-readable descriptions, making semantic search and code understanding less effective. This Skill guides an agent through iteratively writing concise descriptions for every symbol in the index. ## Core Features & Use Cases - Batch Enrichment Loop: Fetches batches of unenriched symbols via `hero_code` with `action: unenriched` and submits descriptions through `hero_enrich` until none remain. - Description Guidelines: Enforces a consistent style—one sentence starting with a verb, focused on behavior and business purpose rather than signatures. - Use Case: After indexing a new repository with `hero scan --code`, run this workflow so every function, struct, and type gains a searchable natural-language description. ## Quick Start Run hero scan --code, then enrich all unenriched symbols in this repository with concise one-line descriptions.

Frequently Asked Questions about deep-code-enrichment

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

FAQPage Schema

How do I add descriptions to symbols in a code index?▼

Run hero scan --code to refresh the index, then call hero_code with action set to unenriched to get symbols needing descriptions. Write a one-line description for each and submit them as a JSON array via hero_enrich, repeating until no symbols remain.

What makes a good code symbol description for search?▼

A good description is one sentence starting with a verb that states what the symbol does, not its signature. Describe behavior and key side effects for functions, and what types represent; skip trivial getters and setters in favor of business purpose.

Do I need to run a code scan before enriching symbols?▼

Yes, run hero scan --code first so the code index is current. Enriching against a stale index can produce descriptions for symbols that have changed or no longer exist.

When is the code enrichment loop finished?▼

The loop is finished when hero_code with action unenriched returns no more symbols. Repeat the fetch, describe, and submit cycle until the unenriched batch is empty.