weclaw-wechat-ai-bridge
WeClaw — WeChat AI Agent Bridge
Skill by ara.so — Daily 2026 Skills collection.
WeClaw connects WeChat to AI agents (Claude, Codex, Gemini, Kimi, OpenClaw, etc.) via a Go-based bridge. It handles QR-code login, message routing, media conversion, and agent lifecycle. Supports three agent modes: ACP (JSON-RPC subprocess, fastest), CLI (new process per message), and HTTP (OpenAI-compatible REST).
Installation
# One-line installer
curl -sSL https://raw.githubusercontent.com/fastclaw-ai/weclaw/main/install.sh | sh
# Via Go toolchain
go install github.com/fastclaw-ai/weclaw@latest
# Via Docker
docker run -it -v ~/.weclaw:/root/.weclaw ghcr.io/fastclaw-ai/weclaw start
First-Run Flow
weclaw start # Shows QR code → scan with WeChat → auto-detects agents → saves config
weclaw login # Add/re-authenticate a WeChat account
weclaw status # Show running state and active agent
weclaw stop # Stop the background daemon
weclaw start -f # Foreground mode (debug/verbose)
Logs: ~/.weclaw/weclaw.log
Config: ~/.weclaw/config.json
Configuration
{
"default_agent": "claude",
"agents": {
"claude": {
"type": "acp",
"command": "/usr/local/bin/claude-agent-acp",
"model": "sonnet"
},
"codex": {
"type": "acp",
"command": "/usr/local/bin/codex-acp"
},
"claude-cli": {
"type": "cli",
"command": "/usr/local/bin/claude",
"args": ["--dangerously-skip-permissions"]
},
"codex-cli": {
"type": "cli",
"command": "/usr/local/bin/codex",
"args": ["--skip-git-repo-check"]
},
"openclaw": {
"type": "http",
"endpoint": "https://api.example.com/v1/chat/completions",
"api_key": "$OPENCLAW_GATEWAY_TOKEN",
"model": "openclaw:main"
}
}
}
Environment Variables
| Variable | Purpose |
|---|---|
WECLAW_DEFAULT_AGENT |
Override default agent at runtime |
WECLAW_API_ADDR |
Change local HTTP API address (default 127.0.0.1:18011) |
OPENCLAW_GATEWAY_URL |
HTTP agent endpoint |
OPENCLAW_GATEWAY_TOKEN |
HTTP agent API token |
Agent Modes
| Mode | Process model | Best for |
|---|---|---|
acp |
Long-running subprocess, JSON-RPC over stdio | Claude, Codex, Kimi, Gemini — fastest, session reuse |
cli |
New process per message, --resume for sessions |
claude -p, codex exec |
http |
OpenAI-compatible /v1/chat/completions |
Any REST-accessible model |
Auto-detection prefers acp over cli when both binaries exist.
Chat Commands (send as WeChat messages)
| Command | Action |
|---|---|
hello |
Send to default agent |
/codex write a sort function |
Route to named agent |
/cc explain this code |
Use alias (/cc → claude) |
/claude |
Switch default agent to Claude (persisted) |
/status |
Show active agent info |
/help |
List available commands |
Built-in Aliases
| Alias | Agent |
|---|---|
/cc |
claude |
/cx |
codex |
/cs |
cursor |
/km |
kimi |
/gm |
gemini |
/ocd |
opencode |
/oc |
openclaw |
Proactive Messaging — CLI
# Send plain text
weclaw send --to "user_id@im.wechat" --text "Hello from WeClaw"
# Send an image
weclaw send --to "user_id@im.wechat" --media "https://example.com/photo.png"
# Send text + media together
weclaw send --to "user_id@im.wechat" \
--text "Check this out" \
--media "https://example.com/photo.png"
# Send a file
weclaw send --to "user_id@im.wechat" --media "https://example.com/report.pdf"
Proactive Messaging — HTTP API
The local API listens on 127.0.0.1:18011 while weclaw start is running.
# Send text
curl -X POST http://127.0.0.1:18011/api/send \
-H "Content-Type: application/json" \
-d '{"to": "user_id@im.wechat", "text": "Hello from WeClaw"}'
# Send image
curl -X POST http://127.0.0.1:18011/api/send \
-H "Content-Type: application/json" \
-d '{"to": "user_id@im.wechat", "media_url": "https://example.com/photo.png"}'
# Send text + media
curl -X POST http://127.0.0.1:18011/api/send \
-H "Content-Type: application/json" \
-d '{"to": "user_id@im.wechat", "text": "See this", "media_url": "https://example.com/photo.png"}'
Supported media types: png, jpg, gif, webp, mp4, mov, pdf, doc, zip.
Change listen address:
WECLAW_API_ADDR=0.0.0.0:18011 weclaw start
Go Integration Example
Call the WeClaw HTTP API from a Go service to send notifications:
package main
import (
"bytes"
"encoding/json"
"fmt"
"net/http"
"os"
)
type SendRequest struct {
To string `json:"to"`
Text string `json:"text,omitempty"`
MediaURL string `json:"media_url,omitempty"`
}
type SendResponse struct {
OK bool `json:"ok"`
Message string `json:"message,omitempty"`
}
func sendToWeChat(to, text, mediaURL string) error {
apiAddr := os.Getenv("WECLAW_API_ADDR")
if apiAddr == "" {
apiAddr = "127.0.0.1:18011"
}
req := SendRequest{To: to, Text: text, MediaURL: mediaURL}
body, err := json.Marshal(req)
if err != nil {
return fmt.Errorf("marshal: %w", err)
}
resp, err := http.Post(
fmt.Sprintf("http://%s/api/send", apiAddr),
"application/json",
bytes.NewReader(body),
)
if err != nil {
return fmt.Errorf("post: %w", err)
}
defer resp.Body.Close()
var result SendResponse
if err := json.NewDecoder(resp.Body).Decode(&result); err != nil {
return fmt.Errorf("decode: %w", err)
}
if !result.OK {
return fmt.Errorf("weclaw error: %s", result.Message)
}
return nil
}
func main() {
recipient := os.Getenv("WECHAT_RECIPIENT_ID") // e.g. "user_id@im.wechat"
if err := sendToWeChat(recipient, "Build succeeded ✅", ""); err != nil {
fmt.Fprintf(os.Stderr, "failed: %v\n", err)
os.Exit(1)
}
fmt.Println("Message sent.")
}
Docker Setup
# Build image
docker build -t weclaw .
# Step 1: Interactive login (scan QR code)
docker run -it -v ~/.weclaw:/root/.weclaw weclaw login
# Step 2: Run daemon with HTTP agent
docker run -d --name weclaw \
-v ~/.weclaw:/root/.weclaw \
-e OPENCLAW_GATEWAY_URL=https://api.example.com \
-e OPENCLAW_GATEWAY_TOKEN="$OPENCLAW_GATEWAY_TOKEN" \
weclaw
# Expose the local API externally (bind carefully — no auth by default)
docker run -d --name weclaw \
-v ~/.weclaw:/root/.weclaw \
-e WECLAW_API_ADDR=0.0.0.0:18011 \
-p 18011:18011 \
-e OPENCLAW_GATEWAY_TOKEN="$OPENCLAW_GATEWAY_TOKEN" \
weclaw
docker logs -f weclaw
ACP/CLI agents require the agent binary inside the container. Mount the binary or build a custom image. HTTP agents work out of the box.
System Service (Auto-start)
macOS (launchd):
cp service/com.fastclaw.weclaw.plist ~/Library/LaunchAgents/
launchctl load ~/Library/LaunchAgents/com.fastclaw.weclaw.plist
Linux (systemd):
sudo cp service/weclaw.service /etc/systemd/system/
sudo systemctl enable --now weclaw
journalctl -u weclaw -f
Development
make dev # Hot reload
go build -o weclaw . # Build binary
./weclaw start -f # Run in foreground
Releasing:
git tag v0.1.0
git push origin v0.1.0
# GitHub Actions builds darwin/linux × amd64/arm64 and uploads release artifacts
Common Patterns
Pattern: Per-user agent routing
Send /claude or /codex as a WeChat message to switch the default agent. The choice persists in ~/.weclaw/config.json across restarts.
Pattern: CI/CD build notifications
After a build, call weclaw send or POST to the HTTP API to push results to a WeChat contact or group.
Pattern: Media from agent
If an agent reply contains , WeClaw auto-downloads, AES-128-ECB encrypts, uploads to WeChat CDN, and delivers as a native image message — no extra config needed.
Pattern: Disable permission prompts for headless use
{
"claude": { "type": "cli", "command": "/usr/local/bin/claude",
"args": ["--dangerously-skip-permissions"] },
"codex": { "type": "cli", "command": "/usr/local/bin/codex",
"args": ["--skip-git-repo-check"] }
}
ACP agents handle permissions automatically and do not need these flags.
Troubleshooting
| Symptom | Fix |
|---|---|
| QR code not appearing | Run weclaw login explicitly; ensure terminal supports UTF-8 |
| Agent not auto-detected | Check binary is on $PATH; run weclaw status |
connection refused on HTTP API |
Confirm weclaw start is running; check WECLAW_API_ADDR |
| Agent permission prompts block responses | Add --dangerously-skip-permissions (Claude) or --skip-git-repo-check (Codex) to args; or use ACP mode |
| Docker — no agent binary | Mount binary: -v /usr/local/bin/claude:/usr/local/bin/claude; or use HTTP mode |
| Markdown not rendering | WeClaw strips markdown automatically for WeChat plain-text display; this is expected |
| Logs | tail -f ~/.weclaw/weclaw.log or docker logs -f weclaw |