@seekrit/mcp 0.1.0
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 +67 -0
- package/dist/index.js +2230 -0
- package/package.json +32 -0
package/README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
1
|
+
# @seekrit/mcp
|
|
2
|
+
|
|
3
|
+
The [seekrit](https://seekrit.dev) MCP server, as a standalone `npx`-able
|
|
4
|
+
package — so an AI agent (Claude Code, or any [MCP](https://modelcontextprotocol.io)
|
|
5
|
+
client) can provision, manage, and inject end-to-end encrypted secrets with
|
|
6
|
+
**zero prior install**.
|
|
7
|
+
|
|
8
|
+
seekrit is zero-knowledge: secret values, data keys, and private keys never
|
|
9
|
+
reach the server. Decryption only ever happens where the credential lives, so
|
|
10
|
+
this MCP server runs **on your machine**, next to the credential — everything
|
|
11
|
+
that produces plaintext (a secret value, a data key, a decryption-capable grant)
|
|
12
|
+
stays on the client. It is the *crypto plane* of seekrit's two-server design.
|
|
13
|
+
|
|
14
|
+
This package is a thin entrypoint around the exact same server as the
|
|
15
|
+
`seekrit mcp` subcommand of [`@seekrit/cli`](https://www.npmjs.com/package/@seekrit/cli);
|
|
16
|
+
it just publishes it as its own binary so no CLI install is required first.
|
|
17
|
+
|
|
18
|
+
## Setup
|
|
19
|
+
|
|
20
|
+
Add it to any MCP client with a stdio server whose command runs this package via
|
|
21
|
+
`npx`, passing a credential in its environment. For Claude Code:
|
|
22
|
+
|
|
23
|
+
```sh
|
|
24
|
+
# An admin token lets the agent provision structure (apps/envs/tokens) too:
|
|
25
|
+
claude mcp add seekrit --env SEEKRIT_TOKEN=skt_… -- npx -y @seekrit/mcp
|
|
26
|
+
```
|
|
27
|
+
|
|
28
|
+
Or drop a `.mcp.json` in your project — the whole setup is this snippet:
|
|
29
|
+
|
|
30
|
+
```json
|
|
31
|
+
{
|
|
32
|
+
"mcpServers": {
|
|
33
|
+
"seekrit": {
|
|
34
|
+
"command": "npx",
|
|
35
|
+
"args": ["-y", "@seekrit/mcp"],
|
|
36
|
+
"env": {
|
|
37
|
+
"SEEKRIT_TOKEN": "skt_…"
|
|
38
|
+
}
|
|
39
|
+
}
|
|
40
|
+
}
|
|
41
|
+
}
|
|
42
|
+
```
|
|
43
|
+
|
|
44
|
+
Requires Node ≥ 20. Set `SEEKRIT_API_URL` too if you're not on the hosted API.
|
|
45
|
+
|
|
46
|
+
## Choosing the credential
|
|
47
|
+
|
|
48
|
+
The server authenticates exactly like the CLI — `SEEKRIT_TOKEN`, or saved
|
|
49
|
+
config at `~/.config/seekrit/config.json`.
|
|
50
|
+
|
|
51
|
+
| Credential | Good for | Notes |
|
|
52
|
+
| --- | --- | --- |
|
|
53
|
+
| **Admin token** (`seekrit token create --admin`) | Provisioning: create apps/groups/envs, compose, grant, mint tokens | Org-scoped; the only headless way to create structure. |
|
|
54
|
+
| **Runtime token** (bound to an env) | Reading/writing/injecting one environment's secrets | Self-decrypts — no passphrase. Cannot provision. |
|
|
55
|
+
|
|
56
|
+
Under user auth (no token), tools that decrypt need `SEEKRIT_PASSPHRASE` in the
|
|
57
|
+
server's environment — there is no TTY to prompt on.
|
|
58
|
+
|
|
59
|
+
## Using secrets without exposing them
|
|
60
|
+
|
|
61
|
+
Prefer the **`run_command`** tool: it injects the resolved secrets into a child
|
|
62
|
+
process and returns only its output, so plaintext never enters the agent's
|
|
63
|
+
context. `get_secret` returns metadata by default and only decrypts when you
|
|
64
|
+
pass `reveal: true`.
|
|
65
|
+
|
|
66
|
+
See the [AI agents guide](https://seekrit.dev/docs/guides/ai-agents) for the
|
|
67
|
+
full tool list and a typical session.
|