Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Deepclaude Proxy

CSecurity

Use Claude Code's autonomous agent loop with DeepSeek V4 Pro, OpenRouter, or any Anthropic-compatible backend at up to 17x lower cost.

81 stars
0 votes
0 copies
0 views
Added 9/19/2026
ai-agentsshellbashnoderefactoringgitapibackend

Works with

claude codecursorvscodeterminalcliapimcp

Security Analysis

C63/100
mediumUses curl or wget to download content
criticalModifies startup scripts or system services for persistence
criticalModifies startup scripts or system services for persistence
mediumInstalls packages at runtime which could introduce malicious dependencies

Scanned 9/19/2026

Install to Claude Code

$npx -y skills add reason-machines/trending-skills --skill deepclaude-proxy --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Deepclaude Proxy?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Deepclaude Proxy
[![Security: C — Skills Directory](https://www.skillsdirectory.com/api/skills/reason-machines-deepclaude-proxy/badge)](https://www.skillsdirectory.com/skills/reason-machines-deepclaude-proxy)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: deepclaude-proxy
description: Use Claude Code's autonomous agent loop with DeepSeek V4 Pro, OpenRouter, or any Anthropic-compatible backend at up to 17x lower cost.
triggers:
  - set up deepclaude with deepseek
  - use claude code with deepseek backend
  - cheaper alternative to claude code
  - switch claude code to openrouter
  - proxy claude code api calls to deepseek
  - deepclaude installation and configuration
  - save money on claude code usage
  - live switch between anthropic and deepseek
---

# deepclaude

> Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection.

**deepclaude** routes Claude Code's API calls to DeepSeek V4 Pro, OpenRouter, Fireworks AI, or any Anthropic-compatible backend — keeping the full Claude Code UX (file editing, bash, subagents, git) while cutting costs by up to 17x.

## How it works

Claude Code reads specific environment variables to determine its API endpoint and model names. `deepclaude` sets these per-session (not permanently), launches Claude Code, then restores originals on exit.

```
Your terminal
  └── Claude Code CLI (tool loop, file editing, bash, git — unchanged)
        └── API calls → DeepSeek V4 Pro ($0.87/M) instead of Anthropic ($15/M)
```

Key environment variables Claude Code uses:

| Variable | Purpose |
|---|---|
| `ANTHROPIC_BASE_URL` | API endpoint override |
| `ANTHROPIC_AUTH_TOKEN` | API key for the backend |
| `ANTHROPIC_DEFAULT_OPUS_MODEL` | Model for Opus-tier tasks |
| `ANTHROPIC_DEFAULT_SONNET_MODEL` | Model for Sonnet-tier tasks |
| `ANTHROPIC_DEFAULT_HAIKU_MODEL` | Model for Haiku-tier / subagents |
| `CLAUDE_CODE_SUBAGENT_MODEL` | Model for spawned subagents |

## Installation

### Prerequisites
- Claude Code CLI installed (`npm install -g @anthropic-ai/claude-code`)
- Node.js 18+ (for the proxy server)
- API key for at least one backend

### 1. Clone the repo

```bash
git clone https://github.com/aattaran/deepclaude.git
cd deepclaude
```

### 2. Set API keys

**macOS/Linux:**
```bash
# DeepSeek (default backend — get key at platform.deepseek.com)
echo 'export DEEPSEEK_API_KEY="$DEEPSEEK_API_KEY"' >> ~/.bashrc

# OpenRouter (optional — cheapest US latency)
echo 'export OPENROUTER_API_KEY="$OPENROUTER_API_KEY"' >> ~/.bashrc

# Fireworks AI (optional — fastest inference)
echo 'export FIREWORKS_API_KEY="$FIREWORKS_API_KEY"' >> ~/.bashrc

source ~/.bashrc
```

**Windows (PowerShell):**
```powershell
setx DEEPSEEK_API_KEY $env:DEEPSEEK_API_KEY
setx OPENROUTER_API_KEY $env:OPENROUTER_API_KEY
setx FIREWORKS_API_KEY $env:FIREWORKS_API_KEY
```

### 3. Install the CLI

**macOS/Linux:**
```bash
chmod +x deepclaude.sh
sudo ln -s "$(pwd)/deepclaude.sh" /usr/local/bin/deepclaude
```

**Windows (PowerShell):**
```powershell
Copy-Item deepclaude.ps1 "$env:USERPROFILE\.local\bin\deepclaude.ps1"
# Or add repo directory to PATH:
setx PATH "$env:PATH;C:\path\to\deepclaude"
```

## Key CLI commands

```bash
# Launch Claude Code with DeepSeek V4 Pro (default)
deepclaude

# Show available backends and configured API keys
deepclaude --status

# Select a specific backend
deepclaude --backend ds          # DeepSeek (default)
deepclaude --backend or          # OpenRouter
deepclaude --backend fw          # Fireworks AI
deepclaude --backend anthropic   # Normal Claude Opus

# Show pricing comparison
deepclaude --cost

# Latency benchmark across all configured providers
deepclaude --benchmark

# Switch backend mid-session (proxy must be running)
deepclaude --switch ds
deepclaude --switch or
deepclaude --switch anthropic

# Remote control — open session in any browser
deepclaude --remote
deepclaude --remote --backend or
deepclaude --remote --backend anthropic
```

## Supported backends

| Backend | Flag | Input/M | Output/M | Notes |
|---|---|---|---|---|
| DeepSeek | `ds` | $0.44 | $0.87 | Auto context caching (120x cheaper on repeat turns) |
| OpenRouter | `or` | $0.44 | $0.87 | Lowest latency from US/EU |
| Fireworks AI | `fw` | $1.74 | $3.48 | Fastest inference, US servers |
| Anthropic | `anthropic` | $3.00 | $15.00 | Original Claude Opus |

## Live backend switching (no restart)

A local proxy runs on `localhost:3200` and intercepts all API calls. Switch backends instantly without restarting Claude Code.

### Proxy control endpoints

```bash
# Switch backend
curl -sX POST http://127.0.0.1:3200/_proxy/mode -d "backend=deepseek"
curl -sX POST http://127.0.0.1:3200/_proxy/mode -d "backend=openrouter"
curl -sX POST http://127.0.0.1:3200/_proxy/mode -d "backend=anthropic"

# Check current backend + uptime
curl -s http://127.0.0.1:3200/_proxy/status

# Token usage and cost savings vs Anthropic
curl -s http://127.0.0.1:3200/_proxy/cost
```

Cost endpoint response shape:
```json
{
  "backends": {
    "deepseek": {
      "input_tokens": 125000,
      "output_tokens": 45000,
      "requests": 12,
      "cost": 0.0941,
      "anthropic_equivalent": 1.05
    }
  },
  "total_cost": 0.0941,
  "anthropic_equivalent": 1.05,
  "savings": 0.9559
}
```

### Slash commands (recommended UX)

Create files in `~/.claude/commands/` to switch backends from inside any Claude Code session:

**`~/.claude/commands/deepseek.md`:**
```markdown
Switch the model proxy to DeepSeek. Run this command silently and report the result:
curl -sX POST http://127.0.0.1:3200/_proxy/mode -d "backend=deepseek"
If successful, say: "Switched to DeepSeek."
```

**`~/.claude/commands/anthropic.md`:**
```markdown
Switch the model proxy back to Anthropic. Run this command silently and report the result:
curl -sX POST http://127.0.0.1:3200/_proxy/mode -d "backend=anthropic"
If successful, say: "Switched to Anthropic."
```

**`~/.claude/commands/openrouter.md`:**
```markdown
Switch the model proxy to OpenRouter. Run this command silently and report the result:
curl -sX POST http://127.0.0.1:3200/_proxy/mode -d "backend=openrouter"
If successful, say: "Switched to OpenRouter."
```

Then type `/deepseek`, `/anthropic`, or `/openrouter` in any Claude Code session.

### VS Code keyboard shortcuts

**`.vscode/tasks.json`:**
```json
{
  "version": "2.0.0",
  "tasks": [
    {
      "label": "Proxy: Switch to DeepSeek",
      "type": "shell",
      "command": "curl -sX POST http://127.0.0.1:3200/_proxy/mode -d 'backend=deepseek'",
      "presentation": { "reveal": "always" },
      "problemMatcher": []
    },
    {
      "label": "Proxy: Switch to Anthropic",
      "type": "shell",
      "command": "curl -sX POST http://127.0.0.1:3200/_proxy/mode -d 'backend=anthropic'",
      "presentation": { "reveal": "always" },
      "problemMatcher": []
    },
    {
      "label": "Proxy: Switch to OpenRouter",
      "type": "shell",
      "command": "curl -sX POST http://127.0.0.1:3200/_proxy/mode -d 'backend=openrouter'",
      "presentation": { "reveal": "always" },
      "problemMatcher": []
    }
  ]
}
```

**`keybindings.json`:**
```json
[
  { "key": "ctrl+alt+d", "command": "workbench.action.tasks.runTask", "args": "Proxy: Switch to DeepSeek" },
  { "key": "ctrl+alt+a", "command": "workbench.action.tasks.runTask", "args": "Proxy: Switch to Anthropic" },
  { "key": "ctrl+alt+o", "command": "workbench.action.tasks.runTask", "args": "Proxy: Switch to OpenRouter" }
]
```

## VS Code / Cursor terminal profiles

**`settings.json` on macOS/Linux:**
```json
{
  "terminal.integrated.profiles.linux": {
    "DeepSeek Agent": {
      "path": "/usr/local/bin/deepclaude"
    },
    "DeepSeek (OpenRouter)": {
      "path": "/usr/local/bin/deepclaude",
      "args": ["--backend", "or"]
    }
  },
  "terminal.integrated.defaultProfile.linux": "DeepSeek Agent"
}
```

**`settings.json` on Windows:**
```json
{
  "terminal.integrated.profiles.windows": {
    "DeepSeek Agent": {
      "path": "powershell.exe",
      "args": ["-ExecutionPolicy", "Bypass", "-NoExit", "-File", "C:\\path\\to\\deepclaude.ps1"]
    }
  }
}
```

## Remote control (browser UI)

Open a Claude Code session in any browser with DeepSeek as the backend:

```bash
deepclaude --remote                  # DeepSeek backend
deepclaude --remote --backend or     # OpenRouter backend
deepclaude --remote --backend anthropic  # Normal Anthropic
```

Traffic split:
```
claude remote-control
  ├── Bridge WebSocket → wss://bridge.claudeusercontent.com (Anthropic, required)
  └── Model API calls  → http://localhost:3200 (proxy)
                          ├── /v1/messages → active backend (DeepSeek/OR/etc.)
                          └── everything else → Anthropic (passthrough)
```

**Requirements for remote control:**
- Must be logged in: `claude auth login`
- Must have a claude.ai subscription (bridge is Anthropic infrastructure)

## Manually setting environment variables (without the CLI wrapper)

If you want to wire up the backend yourself without using the `deepclaude` wrapper script:

```bash
# DeepSeek
export ANTHROPIC_BASE_URL="https://api.deepseek.com/v1"
export ANTHROPIC_AUTH_TOKEN="$DEEPSEEK_API_KEY"
export ANTHROPIC_DEFAULT_OPUS_MODEL="deepseek-chat"
export ANTHROPIC_DEFAULT_SONNET_MODEL="deepseek-chat"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="deepseek-chat"
export CLAUDE_CODE_SUBAGENT_MODEL="deepseek-chat"
claude  # launch normally

# OpenRouter
export ANTHROPIC_BASE_URL="https://openrouter.ai/api/v1"
export ANTHROPIC_AUTH_TOKEN="$OPENROUTER_API_KEY"
export ANTHROPIC_DEFAULT_OPUS_MODEL="deepseek/deepseek-chat-v3-0324"
export ANTHROPIC_DEFAULT_SONNET_MODEL="deepseek/deepseek-chat-v3-0324"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="deepseek/deepseek-chat-v3-0324"
export CLAUDE_CODE_SUBAGENT_MODEL="deepseek/deepseek-chat-v3-0324"
claude
```

## What works and what doesn't

### ✅ Fully supported
- File reading, writing, editing (`Read`/`Write`/`Edit` tools)
- Bash / PowerShell execution
- Glob and Grep search
- Multi-step autonomous tool loops
- Subagent spawning
- Git operations
- Project initialization (`/init`)
- Thinking mode (enabled by default)

### ⚠️ Not supported or degraded

| Feature | Reason |
|---|---|
| Image / vision input | DeepSeek's Anthropic-compat endpoint doesn't support images |
| MCP server tools | Not supported through compatibility layer |
| Anthropic prompt caching (`cache_control`) | Ignored; DeepSeek uses its own automatic caching |
| Parallel tool use | DeepSeek supports up to 128/call but Claude Code sends sequentially |

## Choosing the right backend

| Task type | Recommendation |
|---|---|
| Routine coding, refactoring, CRUD | DeepSeek (default) — 90% cost savings |
| US/EU latency sensitive | OpenRouter |
| Highest throughput / speed | Fireworks AI |
| Complex multi-step reasoning, architecture | Switch to `--backend anthropic` |

Use `/deepseek` for most work, `/anthropic` for the hard 20% — then switch back.

## Troubleshooting

**`DEEPSEEK_API_KEY` not found:**
```bash
# Verify the var is set in current shell
echo $DEEPSEEK_API_KEY
# If blank, re-source your profile
source ~/.bashrc   # or ~/.zshrc
```

**`deepclaude: command not found`:**
```bash
# Check symlink
ls -la /usr/local/bin/deepclaude
# Re-create if missing
sudo ln -sf "$(pwd)/deepclaude.sh" /usr/local/bin/deepclaude
```

**Proxy not responding on port 3200:**
```bash
# Check if proxy process is running
lsof -i :3200
# Restart deepclaude — proxy starts automatically
deepclaude
```

**`curl` not available on Windows for slash commands:**
```powershell
# Use PowerShell equivalent in the .md command files:
Invoke-RestMethod -Uri http://127.0.0.1:3200/_proxy/mode -Method Post -Body 'backend=deepseek'
```

**Image inputs fail with DeepSeek:**
- Switch to Anthropic for that session: `deepclaude --switch anthropic`
- Or launch with: `deepclaude --backend anthropic`

**Remote control fails:**
```bash
# Ensure you're authenticated
claude auth login
# Verify subscription at claude.ai — bridge requires it
```

Attribution

reason-machinesreason-machines
View sourceMore from reason-machines →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.

1023331 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3331 votes

catchup

Recovers prior coding-agent session context by running `catchup <agent> --since-compact`, which extracts a clean summary of a previous Codex, Claude Code, Antigravity, OpenCode, or Pi Agent session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", or asks to recover/summarize a previous session before continuing. Do NOT use for the current conversation, git history, or any non-agent log.

611 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →