Aller au contenu
TakoAPI
Retour aux skills

context-builder

par igorls

Coding Agents & IDEs
500Téléchargements rapportés par clawskills.sh
0Vues
0 mention J'aimeConnecte-toi pour aimer ce skill
—Pas encore noté
Ta note

Résumé

Generate LLM-optimized codebase context from any directory using context-builder CLI.

Description

Ce que fait ce skill

context-builder scans a local directory and generates a single structured markdown file from the codebase. Output is ordered by file relevance and supports AST-aware signature extraction to minimize token usage. Token budgeting, .gitignore awareness, and automatic directory exclusions are built in.

AST-aware signatures reduce context size by 80-90% compared to raw file dumps, making large codebases fit within model token limits without manual curation.

Quand l'utiliser

  • Feed an entire Rust codebase to an LLM for architecture review
  • Onboard to an unfamiliar repo by generating a full project snapshot
  • Extract only public API signatures before a code review session
  • Generate incremental diffs after code changes to update LLM context
  • Package a specific monorepo package for focused LLM analysis

Exemple de workflow

ENTRÉE

User asks: Generate a context file for my Rust project at /home/user/myapp for LLM code review

AGENT
  1. 1

    Run token count preview: context-builder -d /home/user/myapp --token-count

  2. 2

    If over budget, scope to Rust and TOML files with a token cap

  3. 3

    Run: context-builder -d /home/user/myapp -f rs,toml --max-tokens 120000 -y -o docs/context.md

  4. 4

    Attach docs/context.md to the LLM prompt for architecture or bug analysis

SORTIE

docs/context.md containing relevance-ordered Rust and TOML files, capped at 120K tokens, ready for LLM input

Prérequis

  • Rust toolchain (rustup + cargo, required to build context-builder from source)

Installation

clawhub install igorls/context-builder

Exécute la commande CLI, ou colle l'instruction dans Claude Code, Cursor, Codex ou l'agent de code que tu utilises.

Liens