@rag-forge/mcp 0.1.0 → 0.1.2

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.
Files changed (3) hide show
  1. package/LICENSE +21 -0
  2. package/README.md +84 -0
  3. package/package.json +17 -14
package/LICENSE ADDED
@@ -0,0 +1,21 @@
1
+ MIT License
2
+
3
+ Copyright (c) 2026 Femi Adedayo
4
+
5
+ Permission is hereby granted, free of charge, to any person obtaining a copy
6
+ of this software and associated documentation files (the "Software"), to deal
7
+ in the Software without restriction, including without limitation the rights
8
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9
+ copies of the Software, and to permit persons to whom the Software is
10
+ furnished to do so, subject to the following conditions:
11
+
12
+ The above copyright notice and this permission notice shall be included in all
13
+ copies or substantial portions of the Software.
14
+
15
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21
+ SOFTWARE.
package/README.md ADDED
@@ -0,0 +1,84 @@
1
+ # @rag-forge/mcp
2
+
3
+ > Model Context Protocol server exposing RAG-Forge pipeline operations as agent-callable tools.
4
+
5
+ `@rag-forge/mcp` is the MCP (Model Context Protocol) server for RAG-Forge. It lets agents — Claude Desktop, Claude Code, Copilot CLI, Cursor, or any MCP-compatible client — query a RAG pipeline, run audits, score against the RAG Maturity Model, and inspect indexed chunks without leaving the conversation.
6
+
7
+ ## Install
8
+
9
+ ```bash
10
+ npm install -g @rag-forge/mcp
11
+ # or
12
+ pnpm add -g @rag-forge/mcp
13
+ ```
14
+
15
+ You also need the Python runtime that powers the actual pipeline:
16
+
17
+ ```bash
18
+ uv pip install rag-forge-core rag-forge-evaluator rag-forge-observability
19
+ ```
20
+
21
+ ## Run as an MCP server
22
+
23
+ ### stdio transport (Claude Desktop, Cursor)
24
+
25
+ Add to your Claude Desktop config (`claude_desktop_config.json`):
26
+
27
+ ```json
28
+ {
29
+ "mcpServers": {
30
+ "rag-forge": {
31
+ "command": "rag-forge-mcp",
32
+ "args": ["--stdio"]
33
+ }
34
+ }
35
+ }
36
+ ```
37
+
38
+ ### HTTP transport (web agents, MCP gateways)
39
+
40
+ ```bash
41
+ rag-forge-mcp --port 3100
42
+ ```
43
+
44
+ Then point your MCP client at `http://localhost:3100/mcp`.
45
+
46
+ ### From the main CLI
47
+
48
+ If you have `@rag-forge/cli` installed, you can also start the MCP server through it:
49
+
50
+ ```bash
51
+ rag-forge serve --mcp --stdio
52
+ # or
53
+ rag-forge serve --mcp --port 3100
54
+ ```
55
+
56
+ ## What the agent can do
57
+
58
+ Once connected, the agent gains these tools:
59
+
60
+ | Tool | Purpose |
61
+ |---|---|
62
+ | `rag_forge_query` | Run a RAG query against the indexed corpus and return retrieved chunks + generated answer |
63
+ | `rag_forge_audit` | Run an evaluation audit against telemetry or a golden set, returning RMM level + metric scores |
64
+ | `rag_forge_cost` | Estimate the cost of a planned audit before running it |
65
+ | `rag_forge_inspect` | Inspect indexed chunks, embedding stats, and retrieval results for a query |
66
+ | `rag_forge_drift_report` | Compare current pipeline state against a saved baseline |
67
+ | `rag_forge_golden_add` | Add a new question/answer pair to the golden set |
68
+ | `rag_forge_assess` | One-shot RMM assessment without running a full audit |
69
+
70
+ All tools return structured JSON the agent can reason about — perfect for agentic debugging loops where the model decides which audit to run, interprets the report, and proposes the next experiment.
71
+
72
+ ## Project context
73
+
74
+ Run from inside a RAG-Forge project directory (one with a `rag-forge.config.ts` at the root). The MCP server reads that config to know which vector store, embedding model, and judge to use.
75
+
76
+ ## Documentation
77
+
78
+ - **Full docs:** [github.com/hallengray/rag-forge](https://github.com/hallengray/rag-forge#readme)
79
+ - **MCP overview:** [github.com/hallengray/rag-forge/blob/main/apps/docs/content/mcp/overview.mdx](https://github.com/hallengray/rag-forge/blob/main/apps/docs/content/mcp/overview.mdx)
80
+ - **Issues:** [github.com/hallengray/rag-forge/issues](https://github.com/hallengray/rag-forge/issues)
81
+
82
+ ## License
83
+
84
+ MIT — Femi Adedayo
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@rag-forge/mcp",
3
- "version": "0.1.0",
3
+ "version": "0.1.2",
4
4
  "description": "MCP server exposing RAG-Forge pipeline operations as agent-callable tools",
5
5
  "author": "Femi Adedayo",
6
6
  "repository": {
@@ -27,19 +27,14 @@
27
27
  "types": "./dist/index.d.ts"
28
28
  }
29
29
  },
30
- "files": ["dist"],
31
- "scripts": {
32
- "build": "tsup",
33
- "dev": "tsup --watch",
34
- "lint": "eslint src/",
35
- "lint:fix": "eslint src/ --fix",
36
- "typecheck": "tsc --noEmit",
37
- "test": "vitest run"
38
- },
30
+ "files": [
31
+ "dist",
32
+ "README.md"
33
+ ],
39
34
  "dependencies": {
40
35
  "@modelcontextprotocol/sdk": "^1.0.0",
41
- "@rag-forge/shared": "workspace:*",
42
- "zod": "^3.23.0"
36
+ "zod": "^3.23.0",
37
+ "@rag-forge/shared": "0.1.2"
43
38
  },
44
39
  "devDependencies": {
45
40
  "@eslint/js": "^9.0.0",
@@ -53,5 +48,13 @@
53
48
  "engines": {
54
49
  "node": ">=20.0.0"
55
50
  },
56
- "license": "MIT"
57
- }
51
+ "license": "MIT",
52
+ "scripts": {
53
+ "build": "tsup",
54
+ "dev": "tsup --watch",
55
+ "lint": "eslint src/",
56
+ "lint:fix": "eslint src/ --fix",
57
+ "typecheck": "tsc --noEmit",
58
+ "test": "vitest run"
59
+ }
60
+ }