Morph Warpgrep logo

Morph Warpgrep

Organization
letta-ai
morph-warpgrep

Integration guide for Morph's WarpGrep (fast agentic code search) and Fast Apply (10,500 tok/s code editing). Use when building coding agents that need fast, accurate code search or need to apply AI-generated edits to code efficiently. Particularly useful for large codebases, deep logic queries, bug tracing, and code path analysis.

Overview

Publisherletta-ai
Repositoryskills
Skill namemorph-warpgrep
Stars
144
Forks
25
Bundled files
1
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by letta-ai on GitHub. Read the source before you install it.

Installation

Install the Morph Warpgrep AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/letta-ai/skills.git /tmp/skills
mkdir -p .claude/skills
cp -r /tmp/skills/tools/morph-warpgrep .claude/skills/morph-warpgrep
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Morph Warpgrep in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Morph Warpgrep on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Morph Warpgrep is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Morph WarpGrep & Fast Apply

Morph provides two tools that significantly improve coding agent performance:

  • WarpGrep: Agentic code search that's 5x faster than regular search, uses parallel tool calls, achieves 0.73 F1 in ~4 steps
  • Fast Apply: Merges AI edits into code at 10,500 tok/s with 98% accuracy (2x faster than search-replace)

Prerequisites

  1. Get a Morph API key from https://www.morphllm.com/dashboard
  2. Set environment variable:
bash
export MORPH_API_KEY="your-api-key"
  1. Install the Morph SDK:
bash
bun add @morphllm/morphsdk
# or
npm install @morphllm/morphsdk
  1. Ensure ripgrep is installed (required for local search):
bash
# macOS
brew install ripgrep

# Ubuntu/Debian  
sudo apt install ripgrep

# Verify installation
rg --version

Quick Test

After setup, run the included test script on any local repository:

bash
# Clone a test repo (or use any existing codebase)
git clone https://github.com/letta-ai/letta-code.git test-repo

# Install SDK
cd test-repo
bun add @morphllm/morphsdk

# Run test script
export MORPH_API_KEY="your-key"
bun ../scripts/test-warpgrep.ts .

Expected output:

======================================================================
MORPH WARPGREP TEST
======================================================================
Repo: .
SDK: @morphllm/morphsdk
======================================================================

| Query                              | Result | Time   | Files |
|------------------------------------|--------|--------|-------|
| Find the main entry point          | ✅     | 5.2s   | 2     |
| Find authentication logic          | ✅     | 4.1s   | 4     |
| Find where configuration is handled | ✅     | 3.8s   | 3     |
| Find error handling patterns       | ✅     | 4.5s   | 5     |

======================================================================
Results: 4 passed, 0 failed
======================================================================

When to Use

Use WarpGrep When:

  • Searching large codebases (1000+ files)
  • Deep logic queries: bug tracing, code paths, control flow analysis
  • Need to find relevant context without polluting the context window
  • Regular grep returns too many irrelevant results

Use Fast Apply When:

  • Applying AI-generated code edits to existing files
  • Need reliable edit merging (98% accuracy vs ~70% for search-replace)
  • Working with large files where diff formats fail

Don't Use When:

  • Simple exact-match searches (regular grep/rg is free and fast enough)
  • Surface-level queries where semantic search suffices
  • Cost is a major concern (Morph API has usage costs)

Quick Start: WarpGrep

Basic Usage

typescript
import { MorphClient } from '@morphllm/morphsdk';

const morph = new MorphClient({ apiKey: process.env.MORPH_API_KEY });

const result = await morph.warpGrep.execute({
  query: 'Find authentication middleware',
  repoRoot: '.'
});

if (result.success) {
  for (const ctx of result.contexts) {
    console.log(`File: ${ctx.file}`);
    console.log(ctx.content);
  }
} else {
  console.error('Search failed');
}

Response Format

typescript
interface WarpGrepResult {
  success: boolean;
  contexts: Array<{
    file: string;    // File path relative to repo root
    content: string; // File content with relevant code
  }>;
  summary?: string;  // Human-readable summary
}

Using as an Agent Tool

typescript
import { MorphClient } from '@morphllm/morphsdk';
import Anthropic from '@anthropic-ai/sdk';

const morph = new MorphClient({ apiKey: process.env.MORPH_API_KEY });
const anthropic = new Anthropic();

// Define WarpGrep as a tool
const tools = [{
  name: 'warpgrep_search',
  description: 'Search codebase for relevant code. Use for finding implementations, tracing bugs, or understanding code flow.',
  input_schema: {
    type: 'object',
    properties: {
      query: { type: 'string', description: 'What to search for' }
    },
    required: ['query']
  }
}];

// Handle tool calls
async function handleToolCall(name: string, input: { query: string }) {
  if (name === 'warpgrep_search') {
    const result = await morph.warpGrep.execute({
      query: input.query,
      repoRoot: process.cwd()
    });
    
    if (result.success) {
      return result.contexts.map(c => `## ${c.file}\n${c.content}`).join('\n\n');
    }
    return 'No results found';
  }
}

Quick Start: Fast Apply

Fast Apply merges AI-generated edits into existing code:

typescript
import { MorphClient } from '@morphllm/morphsdk';

const morph = new MorphClient({ apiKey: process.env.MORPH_API_KEY });

const result = await morph.fastApply.apply({
  originalCode: `function divide(a, b) {
  return a / b;
}`,
  editSnippet: `function divide(a, b) {
  if (b === 0) throw new Error("Division by zero");
  return a / b;
}`
});

console.log(result.mergedCode);

Direct API (Alternative)

typescript
const response = await fetch('https://api.morphllm.com/v1/chat/completions', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'Authorization': `Bearer ${process.env.MORPH_API_KEY}`
  },
  body: JSON.stringify({
    model: 'morph-v3-fast',  // or 'morph-v3-large' for complex edits
    messages: [{
      role: 'user',
      content: `<instruction>Add error handling</instruction>
<code>function divide(a, b) { return a / b; }</code>
<update>function divide(a, b) {
  if (b === 0) throw new Error("Division by zero");
  return a / b;
}</update>`
    }],
    temperature: 0
  })
});

const data = await response.json();
const mergedCode = data.choices[0].message.content;

Tested Results

Tested on the letta-code repository (~300 TypeScript files) using SDK v0.2.103:

QueryResultTimeFiles Found
"Find authentication logic"4.2ssrc/auth/oauth.ts, src/auth/setup.ts, +2
"Find the main CLI entry point"5.8ssrc/index.ts, src/cli/App.tsx
"Find where models are configured"3.1ssrc/agent/model.ts, src/models.json, +1
"Find how memory blocks work"3.9ssrc/agent/memory.ts, src/agent/memoryFilesystem.ts, +1
"Find the settings manager"2.9ssrc/settings-manager.ts, src/settings.ts

5/5 tests passed

Performance Summary

  • Average time: 4.0 seconds
  • Token efficiency: 39% fewer input tokens vs manual search
  • Accuracy: Finds relevant code in 2-4 turns

How WarpGrep Works

WarpGrep is an agentic search that runs up to 4 turns:

┌─────────────────────────────────────────────────────────────┐
│  Turn 1: Analyze query, map repo structure, initial search  │
├─────────────────────────────────────────────────────────────┤
│  Turn 2-3: Refine search, read specific files               │
├─────────────────────────────────────────────────────────────┤
│  Turn 4: Return all relevant code locations                 │
└─────────────────────────────────────────────────────────────┘

The SDK handles the multi-turn conversation automatically, executing local tools:

ToolDescriptionImplementation
grepRegex search across filesUses ripgrep (rg)
readRead file contentsLocal filesystem
list_dirShow directory structureLocal filesystem

Architecture

┌──────────────────────────────────────────────────────────────┐
│                    Your Code / Agent                         │
└──────────────────────────┬───────────────────────────────────┘
┌──────────────────────────────────────────────────────────────┐
│              @morphllm/morphsdk                              │
│  ┌─────────────────────────────────────────────────────────┐ │
│  │ 1. Build repo structure                                 │ │
│  │ 2. Send query to Morph API                              │ │
│  │ 3. Execute local tools (grep, read, list_dir)           │ │
│  │ 4. Multi-turn refinement                                │ │
│  │ 5. Return relevant code contexts                        │ │
│  └─────────────────────────────────────────────────────────┘ │
└──────────────────────────┬───────────────────────────────────┘
            ┌──────────────┴──────────────┐
            ▼                             ▼
┌───────────────────────┐    ┌───────────────────────┐
│    Morph API          │    │   Local Filesystem    │
│  (morph-warp-grep-v1) │    │   (ripgrep, fs)       │
└───────────────────────┘    └───────────────────────┘

Morph's Benchmarks

From Morph's SWE-bench evaluation with Claude 4.5 Opus:

MetricWithout WarpGrepWith WarpGrepImprovement
Input Tokens14K9K39% fewer
Agent Turns35.026.026% fewer
Tasks Solved74.4%81.9%10% more

Source: Morph WarpGrep Benchmarks


Common Patterns

Reconnaissance-Then-Action

typescript
import { MorphClient } from '@morphllm/morphsdk';

const morph = new MorphClient({ apiKey: process.env.MORPH_API_KEY });

// 1. Search for relevant code
const result = await morph.warpGrep.execute({
  query: 'Where is the payment processing logic?',
  repoRoot: '.'
});

// 2. Use found contexts to inform next steps
if (result.success) {
  const relevantFiles = result.contexts.map(c => c.file);
  console.log('Found relevant files:', relevantFiles);
  // Now read/edit these specific files
}

Combining WarpGrep + Fast Apply

typescript
import { MorphClient } from '@morphllm/morphsdk';

const morph = new MorphClient({ apiKey: process.env.MORPH_API_KEY });

// 1. Find the code to modify
const search = await morph.warpGrep.execute({
  query: 'Find the user validation function',
  repoRoot: '.'
});

if (search.success && search.contexts.length > 0) {
  const targetFile = search.contexts[0];
  
  // 2. Apply an edit
  const result = await morph.fastApply.apply({
    originalCode: targetFile.content,
    editSnippet: '// Add your modified version here'
  });
  
  console.log(result.mergedCode);
}

MCP Integration

For personal use with Claude Code, Cursor, or other MCP clients:

bash
# Install MCP server
claude mcp add morph --scope user -e MORPH_API_KEY=YOUR_API_KEY --npx -y @morphllm/morphmcp

This adds a warpgrep_codebase_search tool to your MCP client.


Troubleshooting

ripgrep Not Found

bash
# Install ripgrep
brew install ripgrep  # macOS
sudo apt install ripgrep  # Ubuntu/Debian

# Verify
rg --version

API Key Issues

bash
# Verify API key works
curl -X POST https://api.morphllm.com/v1/chat/completions \
  -H "Authorization: Bearer $MORPH_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"morph-v3-fast","messages":[{"role":"user","content":"test"}]}'

SDK Version Issues

Ensure you're using the latest SDK:

bash
bun add @morphllm/morphsdk@latest
# or
npm install @morphllm/morphsdk@latest

Cost Considerations

  • WarpGrep uses 1-4 API calls per search (typically 2-3)
  • Fast Apply uses 1 API call per edit
  • Pricing: $0.80 per 1M tokens (input and output)
  • Monitor usage via Morph Dashboard
  • Use regular grep/ripgrep for simple exact-match searches (free)

Resources

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Morph Warpgrep AI skill do?

Integration guide for Morph's WarpGrep (fast agentic code search) and Fast Apply (10,500 tok/s code editing). Use when building coding agents that need fast, accurate code search or need to apply AI-generated edits to code efficiently. Particularly useful for large codebases, deep logic queries, bug tracing, and code path analysis.

Why use Morph Warpgrep on TypingMind?

Because you install it once and use it with any model. Morph Warpgrep is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Morph Warpgrep in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/letta-ai/skills/tree/main/tools/morph-warpgrep. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Morph Warpgrep?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Morph Warpgrep?

As many as you like. As long as a model supports skills, you can use Morph Warpgrep with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Morph Warpgrep AI skill free?

It is published on GitHub by letta-ai. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇