agentgate 0.1.2 → 0.1.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +50 -10
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -60,21 +60,60 @@ The server runs on port 3050 by default. Set `PORT` environment variable to chan
|
|
|
60
60
|
3. Add service accounts (OAuth or API tokens depending on service)
|
|
61
61
|
4. Create API keys for your agents via CLI
|
|
62
62
|
|
|
63
|
-
##
|
|
63
|
+
## Using with Clawdbot / OpenClaw
|
|
64
64
|
|
|
65
|
-
|
|
65
|
+
> ⚠️ **IMPORTANT:** Do NOT run agentgate on the same machine as your AI agent (Clawdbot, OpenClaw, etc.). If the agent has local filesystem access, it could read the database directly and bypass all security controls. Always run agentgate on a **separate, isolated machine** that agents can only reach over the network.
|
|
66
66
|
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
67
|
+
### Configure Your Agent
|
|
68
|
+
|
|
69
|
+
Add agentgate to your agent's `TOOLS.md`:
|
|
70
|
+
|
|
71
|
+
```markdown
|
|
72
|
+
### agentgate
|
|
73
|
+
- Base URL: `https://your-agentgate-server.com`
|
|
74
|
+
- Bearer token: `rms_your_key_here`
|
|
75
|
+
- **URL pattern:** `/api/{service}/{accountName}/...`
|
|
76
|
+
- **Reads (GET):** Execute immediately
|
|
77
|
+
- **Writes (POST/PUT/DELETE):** Queue for human approval
|
|
78
|
+
|
|
79
|
+
#### Write Queue
|
|
80
|
+
# Submit write request
|
|
81
|
+
POST /api/queue/{service}/{accountName}/submit
|
|
82
|
+
body: { requests: [{method, path, body}], comment: "why" }
|
|
83
|
+
|
|
84
|
+
# Check status
|
|
85
|
+
GET /api/queue/{service}/{accountName}/status/{id}
|
|
86
|
+
```
|
|
87
|
+
|
|
88
|
+
Or include in your agent's system prompt:
|
|
89
|
+
|
|
90
|
+
```
|
|
91
|
+
You have access to agentgate at https://your-server.com
|
|
92
|
+
API key: rms_your_key_here
|
|
93
|
+
|
|
94
|
+
For reads: GET /api/{service}/{account}/path
|
|
95
|
+
For writes: POST to /api/queue/{service}/{account}/submit with {requests, comment}
|
|
96
|
+
|
|
97
|
+
Always include a clear comment explaining your intent for write operations.
|
|
98
|
+
A human will review and approve before execution.
|
|
99
|
+
```
|
|
100
|
+
|
|
101
|
+
### Generate a Skill File
|
|
70
102
|
|
|
71
|
-
|
|
72
|
-
npm run keys create <name>
|
|
103
|
+
agentgate can generate an [AgentSkill](https://docs.openclaw.ai/tools/skills) compatible file:
|
|
73
104
|
|
|
74
|
-
|
|
75
|
-
|
|
105
|
+
```bash
|
|
106
|
+
curl -H "Authorization: Bearer rms_your_key" \
|
|
107
|
+
https://your-server.com/api/skill > SKILL.md
|
|
76
108
|
```
|
|
77
109
|
|
|
110
|
+
|
|
111
|
+
|
|
112
|
+
## API Key Management
|
|
113
|
+
|
|
114
|
+
Create and manage API keys for your agents in the admin UI at `/ui/keys`.
|
|
115
|
+
|
|
116
|
+
|
|
78
117
|
## Usage
|
|
79
118
|
|
|
80
119
|
Agents make requests with the API key in the Authorization header:
|
|
@@ -207,10 +246,11 @@ BASE_URL=https://agentgate.yourdomain.com npm start
|
|
|
207
246
|
## TODO
|
|
208
247
|
|
|
209
248
|
- [ ] Per-agent service access control - different agents can access different services/accounts
|
|
210
|
-
|
|
211
249
|
- [ ] Fine-grained endpoint control per service - whitelist/blacklist individual endpoints (even for read operations)
|
|
212
250
|
|
|
213
251
|
## License
|
|
214
252
|
|
|
215
253
|
ISC
|
|
216
254
|
|
|
255
|
+
|
|
256
|
+
|