# LLM Tldr

> Code analysis tool that gives AI agents exactly the code they need - 95% token savings, 300x faster queries across 16 languages for code-analysis, semantic-search, llm-tools, debugging, refactoring, codebase-understanding, token-optimization, ai-assistance, program-analysis, dependency-tracing

- Skill: `lev-os/llm-tldr` (Agent Skill)
- Install (CLI): `npx skillmds@latest add lev-os/llm-tldr`
- Raw SKILL.md: https://api.skillmd.com/api/skills/lev-os/llm-tldr/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: lev-os (https://skillmd.com/u/lev-os)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/lev-os/llm-tldr

---


# llm-tldr

**AI Agent's Code Intelligence Tool** - Transform massive codebases into actionable insights for AI assistants

## Description

llm-tldr is your AI agent's secret weapon for understanding large codebases. Instead of drowning in 100K+ lines of raw code, get structured analysis that fits in your context window with **95% token savings** and **300x faster queries**.

**Repository:** [parcadei/llm-tldr](https://github.com/parcadei/llm-tldr)
**Language:** Python
**Stars:** 285
**License:** Apache License 2.0

## When to Use This Skill

**Use this skill when your AI agent needs to:**

- Analyze large codebases that exceed token limits
- Debug complex bugs across multiple files
- Understand code dependencies and call graphs
- Find functions by behavior, not just text search
- Generate accurate code changes with full context
- Refactor code safely with impact analysis
- Provide codebase overviews for new team members
- Optimize LLM context usage for coding tasks

## What It Does

**llm-tldr provides 5 layers of code intelligence:**

1. **AST Analysis** - Function/class structure extraction
2. **Call Graphs** - Who calls what, reverse dependencies
3. **Control Flow** - Code execution paths and complexity
4. **Data Flow** - Variable tracing and transformations
5. **Program Slicing** - Minimal code affecting specific lines

**Plus semantic search** - Find code by what it _does_, not what it's called

## How It Works

```bash
# Quick setup for any codebase
pip install llm-tldr
tldr warm /path/to/project    # Index once (~30-60s)
tldr context main --project . # Get function summary (99% token savings)
tldr semantic "validate JWT tokens" . # Natural language search
```

**For AI agents:**

- **Before reading code:** `tldr tree src/` + `tldr structure src/ --lang python`
- **Before editing:** `tldr context function_name --project .`
- **Before refactoring:** `tldr impact function_name .`
- **During debugging:** `tldr slice file.py function 42`

## Why It Matters for AI Agents

**Problem:** AI agents struggle with large codebases

- Token limits force incomplete context
- Raw code dumps overwhelm reasoning
- Missing dependencies cause incorrect changes
- No understanding of code behavior vs. naming

**Solution:** llm-tldr gives agents surgical precision

- **95% token reduction** for function contexts
- **300x faster queries** (100ms vs 30s)
- **Behavior-based search** finds code by purpose
- **Dependency tracing** prevents breaking changes
- **Multi-language support** (16 languages)

## Examples

### Example 1: Debug null pointer issue

**Without llm-tldr:** Read 150-line function manually, trace variables, miss control flow bug

**With llm-tldr:**

```bash
tldr slice src/auth.py login 42
```

**Output:** Only 6 relevant lines showing the bug:

```python
3:   user = db.get_user(username)
7:   if user is None:
12:     raise NotFound
28:  token = create_token(user)  # ← BUG: skipped null check
35:  session.token = token
42:  return session
```

### Example 2: Find authentication code

**Without llm-tldr:** Text search for "auth" finds comments but misses actual validation logic

**With llm-tldr:**

```bash
tldr semantic "validate JWT tokens and check expiration" .
```

**Finds:** `verify_access_token()` even without "JWT" in the name, because call graph reveals its purpose

### Example 3: Safe refactoring

**Without llm-tldr:** Guess which tests to run, risk breaking dependencies

**With llm-tldr:**

```bash
tldr impact login .
tldr change-impact src/auth.py
```

**Output:** Exact list of callers and affected test files

## Quick Reference

### Repository Info

- **Homepage:**
- **Topics:**
- **Open Issues:** 3
- **Last Updated:** 2026-01-13

### Languages

- **Python:** 100.0%

### Recent Releases

No releases available

## Available References

- `references/README.md` - Complete README documentation
- `references/CHANGELOG.md` - Version history and changes
- `references/issues.md` - Recent GitHub issues
- `references/releases.md` - Release notes
- `references/file_structure.md` - Repository structure

## Usage

See README.md for complete usage instructions and examples.

---

**Generated by Skill Seeker** | GitHub Repository Scraper

## Technique Map
- **Role definition** - Clarifies operating scope and prevents ambiguous execution.
- **Context enrichment** - Captures required inputs before actions.
- **Output structuring** - Standardizes deliverables for consistent reuse.
- **Step-by-step workflow** - Reduces errors by making execution order explicit.
- **Edge-case handling** - Documents safe fallbacks when assumptions fail.

## Technique Notes
These techniques improve reliability by making intent, inputs, outputs, and fallback paths explicit. Keep this section concise and additive so existing domain guidance remains primary.

## Prompt Architect Overlay
### Role Definition
You are the prompt-architect-enhanced specialist for lev-find-llm-tldr, responsible for deterministic execution of this skill's guidance while preserving existing workflow and constraints.

### Input Contract
- Required: clear user intent and relevant context for this skill.
- Preferred: repository/project constraints, existing artifacts, and success criteria.
- If context is missing, ask focused questions before proceeding.

### Output Contract
- Provide structured, actionable outputs aligned to this skill's existing format.
- Include assumptions and next steps when appropriate.
- Preserve compatibility with existing sections and related skills.

### Edge Cases & Fallbacks
- If prerequisites are missing, provide a minimal safe path and request missing inputs.
- If scope is ambiguous, narrow to the highest-confidence sub-task.
- If a requested action conflicts with existing constraints, explain and offer compliant alternatives.