stacklit-navigator

Guide AI agents to navigate repositories using stacklit.json index sections.

101|8|Updated Apr 9, 2026
One-click install
npx skills add https://github.com/glincker/stacklit --skill stacklit-navigator
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: stacklit-navigator
Source: https://github.com/glincker/stacklit/tree/main
Command: npx skills add https://github.com/glincker/stacklit --skill stacklit-navigator

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Stacklit-navigator helps AI agents avoid wasting context window tokens on repetitive file exploration by using the prebuilt stacklit.json index as a fast, structured navigation layer.

Core Features & Use Cases

  • Index-first navigation: Directs the agent to start from stacklit.json at the repo root and locate relevant modules without grepping or scanning directories.
  • Module-driven decision making: Leverages module fields like purpose, depends_on/depended_by, exports, file_list, and git activity to understand impact and where to read next.
  • Actionable conventions and safety rails: Uses hints (add_feature, test_command, env vars) and explicit “What not to do” rules to reduce wasted effort and avoid high-blast-radius changes.
  • Optional MCP integration: If stacklit serve is available, instructs the agent to use MCP tools (get_overview, get_module, find_module, get_dependencies, get_hot_files, get_hints) instead of manual JSON reading.

Quick Start

Ask the agent to read stacklit.json first and then choose the most relevant module(s) using purpose, depends_on/depended_by, and the module file_list before opening any source files.

Frequently Asked Questions about stacklit-navigator

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

FAQPage Schema
How do I reduce AI agent context window usage when navigating a codebase?▼

Reduce AI agent context burn by using a prebuilt stacklit.json index as a structured navigation layer. This directs the agent to locate relevant modules via purpose and dependency fields without scanning directories or grepping files.

How does index-first codebase navigation work for feature planning?▼

Index-first navigation works by reading stacklit.json at the repository root. The agent reviews module fields like depends_on, depended_by, and file_list to understand impact and decide where to read next during feature planning.

Can I use MCP tools to analyze dependency graphs in my repository?▼

Yes, if stacklit serve is available, you can use MCP tools like get_overview, get_module, get_dependencies, and get_hints instead of manually reading the JSON index to analyze dependency graphs.

What is the best way to find testing hints and environment variables in a large codebase?▼

The best way to find testing hints is to use the actionable conventions stored in the stacklit.json index. The index exposes hints like test_command and env vars to reduce wasted exploration effort.

Does stacklit-navigator require any external dependencies to work?▼

No external dependencies are required. The navigator operates by instructing the agent to read the stacklit.json file directly, though optional MCP integration is supported when stacklit serve is available.

When should I not use manual JSON reading for codebase exploration?▼

You should avoid manual JSON reading when stacklit serve is available. In that scenario, use MCP tools like find_module and get_hot_files to retrieve module data and navigate the repository more efficiently.