mcpdrift-cli 1.0.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,428 @@
+# mcpdrift 🧪
+
+> Dependable CI/CD regression tester and snapshot assertion tool for Model Context Protocol (MCP) servers.
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Node.js](https://img.shields.io/badge/node-%3E%3D18.0.0-brightgreen.svg)](https://nodejs.org)
+
+`mcpdrift` prevents unintended breaking changes and response drift in MCP servers. It executes defined tool calls over `stdio` via the official `@modelcontextprotocol/sdk`, captures response snapshots, and fails your CI pipeline with clear, colorized line-by-line diffs when unexpected changes occur.
+
+---
+
+## Table of Contents
+
+- [What is mcpdrift?](#what-is-mcpdrift)
+- [Important Concepts & Guarantees](#important-concepts--guarantees)
+- [Quickstart (5-Minute Walkthrough)](#quickstart-5-minute-walkthrough)
+- [Configuration Reference (`mcpdrift.config.json`)](#configuration-reference-mcpdriftconfigjson)
+  - [Multiple Test Cases Per Tool](#multiple-test-cases-per-tool)
+  - [Targeted Field Ignoring (RFC 6901 JSON Pointer)](#targeted-field-ignoring-rfc-6901-json-pointer)
+- [CLI Command Reference](#cli-command-reference)
+  - [`mcpdrift run`](#mcpdrift-run)
+  - [`mcpdrift approve`](#mcpdrift-approve)
+  - [`mcpdrift init`](#mcpdrift-init)
+- [CI/CD Integration](#cicd-integration)
+- [Machine-Readable Reports](#machine-readable-reports)
+- [Migration Guide: Single-Tool to Multi-Case IDs](#migration-guide-single-tool-to-multi-case-ids)
+- [Known Limitations](#known-limitations)
+- [License](#license)
+
+---
+
+## What is mcpdrift?
+
+Model Context Protocol (MCP) servers expose tools to AI agents and LLMs. When developers refactor server code, update dependencies, or adjust prompt/response serialization, tool response schemas can inadvertently drift.
+
+`mcpdrift` acts like Jest/Vitest snapshot testing, but specifically tailored for MCP servers:
+1. Spawns your server as a child process via `serverCommand`.
+2. Connects over `stdio` using `@modelcontextprotocol/sdk`.
+3. Calls each configured tool with deterministic test inputs.
+4. Compares responses against committed baseline snapshots under `.mcpdrift/snapshots/<case-id>.json`.
+5. Exits `0` when identical, or exits `1` with a line-by-line diff if unexpected changes or errors occur.
+
+---
+
+## Important Concepts & Guarantees
+
+- **Differences do not automatically mean bugs**: A detected drift indicates the server response changed from the saved baseline. If the change was deliberate, review the diff and run `mcpdrift approve <case-id>` to accept it.
+- **Passing snapshots do not prove correctness**: A snapshot only proves that the current output matches the baseline. If your baseline contained a defect, `mcpdrift` will preserve that defect until updated.
+- **Dynamic data requires controlled fixtures or narrow ignore rules**: Fields like timestamps, UUIDs, trace tokens, or ephemeral counters will drift every run. Use targeted `ignore` rules (RFC 6901 JSON Pointers) or configure deterministic server test seeds.
+- **MCP tools may have side effects**: Tools that delete files, invoke external APIs, or charge accounts will trigger those actions during testing. **Always point `mcpdrift` to local, sandboxed test servers or read-only mock environments.**
+- **Approval executes tools again**: When you run `mcpdrift approve`, `mcpdrift` re-runs the tool against your server and overwrites the snapshot file with the newly returned output. If the tool fails or returns `isError: true`, approval is automatically aborted and your existing baseline is preserved.
+- **CI checks block PRs by policy**: `mcpdrift run --ci` returns exit code `1` on drift, missing baselines, or tool failures. Configure your repository branch protection rules to require this check before merging.
+
+---
+
+## Quickstart
+
+### A. Consumer Quickstart (For Your MCP Server Project)
+
+In your MCP project directory:
+
+#### 1. Installation
+
+Run directly without installation, or add it to your project:
+
+```bash
+# Run directly:
+npx mcpdrift-cli --help
+
+# Or install as a project devDependency:
+npm install --save-dev mcpdrift-cli
+```
+
+#### 2. Initialize Configuration
+Connect to your running MCP server to inspect its available tools. To prevent unintended execution of tools with side effects, explicit tool selection is required:
+
+```bash
+# Explicitly select tools by name:
+npx mcpdrift-cli init --server "node dist/server.js" --tools getWeather,searchDocs
+
+# Or explicitly include all discovered tools:
+npx mcpdrift-cli init --server "node dist/server.js" --all-tools
+
+# Or generate a standalone starter template without connecting to a server:
+npx mcpdrift-cli init --template
+```
+
+This creates `mcpdrift.config.json` in your project root with inferred parameter placeholders.
+
+#### 3. Review and Customize Inputs
+Open `mcpdrift.config.json` and adjust inputs to provide deterministic, safe test values:
+
+```json
+{
+  "serverCommand": "node dist/server.js",
+  "testCases": [
+    {
+      "id": "weather_addis",
+      "tool": "getWeather",
+      "input": { "city": "Addis Ababa" }
+    },
+    {
+      "id": "docs_pricing",
+      "tool": "searchDocs",
+      "input": { "query": "pricing" }
+    }
+  ]
+}
+```
+
+#### 4. Create Local Baseline Snapshots
+Run tests in local development mode:
+
+```bash
+npx mcpdrift-cli run
+```
+
+This creates baseline response snapshots under `.mcpdrift/snapshots/<case-id>.json`. Commit `.mcpdrift/snapshots/` to Git.
+
+#### 5. Run Strict CI Checks
+In CI pipelines (or with `--ci`), `mcpdrift` asserts that all snapshots exist and match:
+
+```bash
+npx mcpdrift-cli run --ci
+```
+
+
+---
+
+### B. Contributor Quickstart (Within the mcpdrift Repository)
+
+If you are developing or testing `mcpdrift` directly from source:
+
+1. Clone the repository and build:
+   ```bash
+   git clone <repo-url>
+   cd mcpdrift
+   npm ci
+   npm run build
+   ```
+
+2. Test against the bundled fixture server:
+   ```bash
+   node dist/cli.js init --server "node examples/sample-server/server.js" --all-tools --output my-test.config.json
+   node dist/cli.js run -c my-test.config.json
+   ```
+
+3. Run the automated offline test suite:
+   ```bash
+   npm test
+   ```
+
+> **Network Access Note**:
+> The automated test suite (`npm test`, covering diffing, pointer rules, config validation, safety, and CLI execution) is **100% offline and deterministic**; it never calls external APIs or npm registries. Installing `mcpdrift` as a consumer via `npm install` requires normal npm registry network access to fetch declared dependencies.
+
+✨ All MCP regression checks passed!
+```
+
+---
+
+## Configuration Reference (`mcpdrift.config.json`)
+
+```json
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "serverCommand": "node dist/server.js",
+  "timeout": 15000,
+  "snapshotDir": ".mcpdrift/snapshots",
+  "env": {
+    "NODE_ENV": "test"
+  },
+  "testCases": [
+    {
+      "id": "weather_cold",
+      "tool": "getWeather",
+      "input": { "city": "Oslo" },
+      "timeout": 5000,
+      "ignore": [
+        "/meta/timestamp",
+        "/meta/traceId"
+      ]
+    },
+    {
+      "id": "weather_warm",
+      "tool": "getWeather",
+      "input": { "city": "Nairobi" }
+    }
+  ]
+}
+```
+
+### Multiple Test Cases Per Tool
+
+You can test the same tool with varying inputs, edge cases, or error parameters. To avoid snapshot collisions, each test case must specify a unique, filesystem-safe `id`.
+
+- Valid characters for `id`: alphanumeric, `_`, `-`, `.`.
+- Path traversal sequences (`..`, `/`, `\`) and special characters (`< > : " | ? *`) are strictly prohibited and rejected by config validation.
+
+### Targeted Field Ignoring (RFC 6901 JSON Pointer)
+
+When responses contain dynamic values (e.g., timestamps, session IDs, tokens), specify an array of RFC 6901 JSON Pointers in `ignore`:
+
+```json
+{
+  "id": "user_profile",
+  "tool": "getUser",
+  "input": { "userId": 42 },
+  "ignore": [
+    "/lastLoginAt",
+    "/session/token",
+    "/tags/0/createdAt"
+  ]
+}
+```
+
+- **Standards-compliant**: Uses standard JSON Pointer syntax starting with `/`. Tokens with `/` are escaped as `~1`, and `~` as `~0`.
+- **Immutable normalization**: `mcpdrift` normalizes copies of both baseline and actual responses without mutating the original output.
+- **Array order preserved**: Array lengths and index positions are preserved.
+- **Missing path safety**: If an ignored path does not exist in a given response, it is safely skipped without failing.
+- **Meaningful changes still caught**: Any modification to unignored fields will immediately trigger a drift failure.
+
+---
+
+## CLI Command Reference
+
+### `mcpdrift run`
+
+Runs all configured test cases against the MCP server.
+
+```bash
+mcpdrift run [options]
+
+Options:
+  -c, --config <path>       Path to config file (default: "mcpdrift.config.json")
+  --ci                      Strict CI mode: fails if any baseline is missing, never writes snapshots
+  --case <id>               Run only a specific test case by ID
+  -t, --tool <name>         Run all test cases configured for a specific tool
+  --report <path>           Write a machine-readable JSON report
+  --json-report <path>      Alias for --report
+  --timeout <ms>            Override tool execution timeout in milliseconds
+  -h, --help                Display command help
+```
+
+### `mcpdrift approve`
+
+Overwrites baseline snapshot(s) with the latest server response.
+
+```bash
+# Approve a specific test case by ID:
+mcpdrift approve weather_addis
+
+# Approve only a specific test case explicitly:
+mcpdrift approve --case weather_addis
+
+# Approve all test cases for a specific tool:
+mcpdrift approve --tool getWeather
+
+# Approve all test cases in the entire suite:
+mcpdrift approve --all
+```
+
+> **Safety Guarantee**: If the tool call throws an exception, times out, or returns `isError: true`, the approval operation is aborted and your existing baseline snapshot remains untouched.
+
+### `mcpdrift init`
+
+Inspects a running MCP server and generates a starter `mcpdrift.config.json`.
+
+```bash
+# Inspect running server:
+mcpdrift init --server "node dist/server.js"
+
+# Filter to specific tools:
+mcpdrift init --server "node dist/server.js" --tools ping,getWeather
+
+# Overwrite existing config:
+mcpdrift init --server "node dist/server.js" --force
+
+# Generate standalone template without connecting to a server:
+mcpdrift init --template
+```
+
+> **Discovery Safety**: `mcpdrift init` issues only `tools/list` over stdio and **never** calls `tools/call`. Inferred inputs are placeholders and must be reviewed.
+
+---
+
+## CI/CD Integration
+
+`mcpdrift` is designed from the ground up for automated pipelines.
+
+### Automatic CI Detection
+
+`mcpdrift` automatically activates strict CI mode when:
+- Passed the `--ci` CLI flag, OR
+- The `CI=true` or `CI=1` environment variable is detected (standard in GitHub Actions, GitLab CI, CircleCI, Jenkins).
+
+In strict CI mode:
+- Any missing baseline snapshot causes an immediate failure (`exit code 1`).
+- No snapshots are written to disk during the CI run.
+- Changed responses or tool errors fail the build with line-by-line diffs.
+
+### Example GitHub Actions Workflow
+
+Create `.github/workflows/mcpdrift.yml`:
+
+```yaml
+name: MCP Server Regression Tests
+
+on:
+  pull_request:
+    branches: [main, master]
+  push:
+    branches: [main, master]
+
+jobs:
+  test:
+    name: MCP Regression (${{ matrix.os }})
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ubuntu-latest, windows-latest]
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '20'
+          cache: 'npm'
+
+      - name: Install dependencies
+        run: npm ci
+
+      - name: Build server
+        run: npm run build --if-present
+
+      - name: Run strict regression check
+        run: npx mcpdrift-cli run --ci
+```
+
+---
+
+## Machine-Readable Reports
+
+Generate machine-readable JSON reports for CI dashboard integration, test summarizers, or compliance logs:
+
+```bash
+npx mcpdrift-cli run --report .mcpdrift/report.json
+```
+
+
+Sample JSON report output (clean JSON, no ANSI escape codes, no leaked secrets):
+
+```json
+{
+  "timestamp": "2026-09-12T18:35:00.000Z",
+  "serverCommand": "node dist/server.js",
+  "ci": true,
+  "summary": {
+    "total": 2,
+    "passed": 1,
+    "changed": 1,
+    "failed": 0,
+    "baselineCreated": 0,
+    "missingBaseline": 0,
+    "success": false
+  },
+  "results": [
+    {
+      "id": "weather_addis",
+      "tool": "getWeather",
+      "status": "PASS",
+      "durationMs": 12,
+      "snapshotPath": "C:\\repo\\.mcpdrift\\snapshots\\weather_addis.json"
+    },
+    {
+      "id": "docs_pricing",
+      "tool": "searchDocs",
+      "status": "CHANGED",
+      "failureCategory": "SNAPSHOT_MISMATCH",
+      "durationMs": 15,
+      "snapshotPath": "C:\\repo\\.mcpdrift\\snapshots\\docs_pricing.json"
+    }
+  ]
+}
+```
+
+---
+
+## Migration Guide: Single-Tool to Multi-Case IDs
+
+In earlier versions of `mcpdrift`, test cases did not require an `id` property, and snapshots were saved as `.mcpdrift/snapshots/<tool-name>.json`.
+
+### Backward Compatibility
+If your configuration contains only **one** test case per tool name, `mcpdrift` maintains 100% backward compatibility:
+- The test case automatically inherits `id: "<tool-name>"`.
+- Existing snapshots at `.mcpdrift/snapshots/<tool-name>.json` will load and pass without change.
+
+### Upgrading to Multiple Cases
+When you want multiple test cases for the same tool:
+1. Add an explicit, unique `id` to each test case in `mcpdrift.config.json`:
+   ```json
+   {
+     "testCases": [
+       { "id": "getWeather_city_1", "tool": "getWeather", "input": { "city": "Addis Ababa" } },
+       { "id": "getWeather_city_2", "tool": "getWeather", "input": { "city": "Tokyo" } }
+     ]
+   }
+   ```
+2. Rename your existing snapshot file to match the new `id`:
+   - e.g., rename `.mcpdrift/snapshots/getWeather.json` to `.mcpdrift/snapshots/getWeather_city_1.json`.
+3. Run `npx mcpdrift-cli run` to generate the new baseline for `getWeather_city_2.json`.
+4. Commit the new snapshots. `mcpdrift` never silently deletes or ambiguously migrates old snapshot files.
+
+---
+
+## Known Limitations
+
+1. **Stdio Only**: `mcpdrift` currently connects to MCP servers over standard I/O (`stdio`). SSE (Server-Sent Events) and custom HTTP transports are planned for future releases.
+2. **Deterministic Inputs Required**: If your MCP server relies on external third-party live APIs that return uncontrolled live data, responses will drift unless mocked or covered by `ignore` rules.
+3. **Re-execution on Approval**: `mcpdrift approve` re-executes the tool against the active server to capture the latest response rather than re-using the cached run failure.
+
+---
+
+## License
+
+[MIT](LICENSE)

package/dist/approver.d.ts

@@ -0,0 +1,8 @@
+export interface ApproveOptions {
+    config?: string;
+    caseId?: string;
+    tool?: string;
+    all?: boolean;
+}
+export declare function approveTool(target?: string, options?: ApproveOptions): Promise<void>;
+//# sourceMappingURL=approver.d.ts.map

package/dist/approver.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"approver.d.ts","sourceRoot":"","sources":["../src/approver.ts"],"names":[],"mappings":"AASA,MAAM,WAAW,cAAc;IAC7B,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,GAAG,CAAC,EAAE,OAAO,CAAC;CACf;AAED,wBAAsB,WAAW,CAAC,MAAM,CAAC,EAAE,MAAM,EAAE,OAAO,GAAE,cAAmB,GAAG,OAAO,CAAC,IAAI,CAAC,CAwG9F"}

package/dist/approver.js

@@ -0,0 +1,90 @@
+import fs from 'node:fs';
+import path from 'node:path';
+import chalk from 'chalk';
+import { loadConfig, getSnapshotPath } from './config.js';
+import { createMcpSession, executeToolWithTimeout } from './client.js';
+import { formatJson } from './differ.js';
+import { writeAtomic } from './fsUtil.js';
+export async function approveTool(target, options = {}) {
+    const { config, resolvedPath } = loadConfig(options.config);
+    const snapshotDir = path.resolve(path.dirname(resolvedPath), config.snapshotDir || path.join('.mcpdrift', 'snapshots'));
+    if (!fs.existsSync(snapshotDir)) {
+        fs.mkdirSync(snapshotDir, { recursive: true });
+    }
+    let casesToApprove = [];
+    if (options.all) {
+        casesToApprove = config.testCases;
+    }
+    else if (options.caseId) {
+        casesToApprove = config.testCases.filter((tc) => (tc.id || tc.tool) === options.caseId);
+        if (casesToApprove.length === 0) {
+            throw new Error(`Test case with ID "${options.caseId}" was not found in ${resolvedPath}.`);
+        }
+    }
+    else if (options.tool) {
+        casesToApprove = config.testCases.filter((tc) => tc.tool === options.tool);
+        if (casesToApprove.length === 0) {
+            throw new Error(`No test cases found for tool "${options.tool}" in ${resolvedPath}.`);
+        }
+    }
+    else if (target) {
+        // Target can match either case ID or tool name
+        const matchById = config.testCases.filter((tc) => (tc.id || tc.tool) === target);
+        if (matchById.length > 0) {
+            casesToApprove = matchById;
+        }
+        else {
+            const matchByTool = config.testCases.filter((tc) => tc.tool === target);
+            if (matchByTool.length > 0) {
+                casesToApprove = matchByTool;
+            }
+            else {
+                throw new Error(`Target "${target}" matched no test case ID or tool name in ${resolvedPath}.`);
+            }
+        }
+    }
+    else {
+        throw new Error('Please specify a case ID or tool name to approve (e.g. `mcpdrift approve <case-id>`) or pass `--all`.');
+    }
+    console.log(chalk.bold(`\n📝 mcpdrift approve: Updating snapshots`));
+    console.log(chalk.dim(`   Config: ${resolvedPath}`));
+    console.log(chalk.dim(`   Server: ${config.serverCommand}`));
+    console.log(chalk.dim(`   Cases to approve: ${casesToApprove.length}\n`));
+    const session = await createMcpSession(config.serverCommand, config.env, config.timeout || 15000);
+    try {
+        for (const testCase of casesToApprove) {
+            const caseId = testCase.id || testCase.tool;
+            const tool = testCase.tool;
+            const snapshotPath = getSnapshotPath(snapshotDir, caseId);
+            const testTimeout = testCase.timeout || config.timeout || 15000;
+            console.log(chalk.dim(`Calling tool "${tool}" for case "${caseId}" to capture latest response...`));
+            let latestResponse;
+            try {
+                latestResponse = await executeToolWithTimeout(session.client, tool, testCase.input || {}, testTimeout);
+            }
+            catch (err) {
+                console.error(chalk.red.bold(`\n❌ Failed to approve "${caseId}": Tool threw an execution error.`));
+                console.error(chalk.red(`   Error: ${err.message}`));
+                console.error(chalk.yellow(`   Existing baseline snapshot was preserved: ${snapshotPath}\n`));
+                throw new Error(`Approval aborted: Tool "${tool}" threw an execution error.`);
+            }
+            // Check if tool returned error result (isError: true)
+            if (latestResponse !== null &&
+                typeof latestResponse === 'object' &&
+                latestResponse.isError === true) {
+                console.error(chalk.red.bold(`\n❌ Failed to approve "${caseId}": Tool returned an error result (isError: true).`));
+                console.error(chalk.red(`   Response:\n${formatJson(latestResponse)}`));
+                console.error(chalk.yellow(`   Existing baseline snapshot was preserved: ${snapshotPath}\n`));
+                throw new Error(`Approval aborted: Tool "${tool}" returned an error result (isError: true).`);
+            }
+            // Write snapshot atomically
+            writeAtomic(snapshotPath, formatJson(latestResponse));
+            console.log(`${chalk.green.bold('✅ APPROVED:')} Snapshot for "${chalk.bold(caseId)}" updated at ${chalk.cyan(path.relative(process.cwd(), snapshotPath))}\n`);
+        }
+    }
+    finally {
+        await session.close();
+    }
+    console.log(chalk.green.bold('🎉 All requested snapshots have been approved!'));
+}
+//# sourceMappingURL=approver.js.map

package/dist/approver.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"approver.js","sourceRoot":"","sources":["../src/approver.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,SAAS,CAAC;AACzB,OAAO,IAAI,MAAM,WAAW,CAAC;AAC7B,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,UAAU,EAAE,eAAe,EAAE,MAAM,aAAa,CAAC;AAC1D,OAAO,EAAE,gBAAgB,EAAE,sBAAsB,EAAE,MAAM,aAAa,CAAC;AACvE,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AACzC,OAAO,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAU1C,MAAM,CAAC,KAAK,UAAU,WAAW,CAAC,MAAe,EAAE,UAA0B,EAAE;IAC7E,MAAM,EAAE,MAAM,EAAE,YAAY,EAAE,GAAG,UAAU,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IAE5D,MAAM,WAAW,GAAG,IAAI,CAAC,OAAO,CAC9B,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAC1B,MAAM,CAAC,WAAW,IAAI,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE,WAAW,CAAC,CAC1D,CAAC;IACF,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,WAAW,CAAC,EAAE,CAAC;QAChC,EAAE,CAAC,SAAS,CAAC,WAAW,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;IACjD,CAAC;IAED,IAAI,cAAc,GAAuB,EAAE,CAAC;IAE5C,IAAI,OAAO,CAAC,GAAG,EAAE,CAAC;QAChB,cAAc,GAAG,MAAM,CAAC,SAAS,CAAC;IACpC,CAAC;SAAM,IAAI,OAAO,CAAC,MAAM,EAAE,CAAC;QAC1B,cAAc,GAAG,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,IAAI,CAAC,KAAK,OAAO,CAAC,MAAM,CAAC,CAAC;QACxF,IAAI,cAAc,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAChC,MAAM,IAAI,KAAK,CAAC,sBAAsB,OAAO,CAAC,MAAM,sBAAsB,YAAY,GAAG,CAAC,CAAC;QAC7F,CAAC;IACH,CAAC;SAAM,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC;QACxB,cAAc,GAAG,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC,IAAI,KAAK,OAAO,CAAC,IAAI,CAAC,CAAC;QAC3E,IAAI,cAAc,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAChC,MAAM,IAAI,KAAK,CAAC,iCAAiC,OAAO,CAAC,IAAI,QAAQ,YAAY,GAAG,CAAC,CAAC;QACxF,CAAC;IACH,CAAC;SAAM,IAAI,MAAM,EAAE,CAAC;QAClB,+CAA+C;QAC/C,MAAM,SAAS,GAAG,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,IAAI,CAAC,KAAK,MAAM,CAAC,CAAC;QACjF,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACzB,cAAc,GAAG,SAAS,CAAC;QAC7B,CAAC;aAAM,CAAC;YACN,MAAM,WAAW,GAAG,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,CAAC,IAAI,KAAK,MAAM,CAAC,CAAC;YACxE,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC3B,cAAc,GAAG,WAAW,CAAC;YAC/B,CAAC;iBAAM,CAAC;gBACN,MAAM,IAAI,KAAK,CAAC,WAAW,MAAM,6CAA6C,YAAY,GAAG,CAAC,CAAC;YACjG,CAAC;QACH,CAAC;IACH,CAAC;SAAM,CAAC;QACN,MAAM,IAAI,KAAK,CACb,uGAAuG,CACxG,CAAC;IACJ,CAAC;IAED,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,2CAA2C,CAAC,CAAC,CAAC;IACrE,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,cAAc,YAAY,EAAE,CAAC,CAAC,CAAC;IACrD,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,cAAc,MAAM,CAAC,aAAa,EAAE,CAAC,CAAC,CAAC;IAC7D,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,wBAAwB,cAAc,CAAC,MAAM,IAAI,CAAC,CAAC,CAAC;IAE1E,MAAM,OAAO,GAAG,MAAM,gBAAgB,CAAC,MAAM,CAAC,aAAa,EAAE,MAAM,CAAC,GAAG,EAAE,MAAM,CAAC,OAAO,IAAI,KAAK,CAAC,CAAC;IAElG,IAAI,CAAC;QACH,KAAK,MAAM,QAAQ,IAAI,cAAc,EAAE,CAAC;YACtC,MAAM,MAAM,GAAG,QAAQ,CAAC,EAAE,IAAI,QAAQ,CAAC,IAAI,CAAC;YAC5C,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC;YAC3B,MAAM,YAAY,GAAG,eAAe,CAAC,WAAW,EAAE,MAAM,CAAC,CAAC;YAC1D,MAAM,WAAW,GAAG,QAAQ,CAAC,OAAO,IAAI,MAAM,CAAC,OAAO,IAAI,KAAK,CAAC;YAEhE,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,iBAAiB,IAAI,eAAe,MAAM,iCAAiC,CAAC,CAAC,CAAC;YAEpG,IAAI,cAAuB,CAAC;YAC5B,IAAI,CAAC;gBACH,cAAc,GAAG,MAAM,sBAAsB,CAC3C,OAAO,CAAC,MAAM,EACd,IAAI,EACJ,QAAQ,CAAC,KAAK,IAAI,EAAE,EACpB,WAAW,CACZ,CAAC;YACJ,CAAC;YAAC,OAAO,GAAQ,EAAE,CAAC;gBAClB,OAAO,CAAC,KAAK,CACX,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,0BAA0B,MAAM,mCAAmC,CAAC,CACpF,CAAC;gBACF,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,aAAa,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC;gBACrD,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,gDAAgD,YAAY,IAAI,CAAC,CAAC,CAAC;gBAC9F,MAAM,IAAI,KAAK,CAAC,2BAA2B,IAAI,6BAA6B,CAAC,CAAC;YAChF,CAAC;YAED,sDAAsD;YACtD,IACE,cAAc,KAAK,IAAI;gBACvB,OAAO,cAAc,KAAK,QAAQ;gBACjC,cAAsB,CAAC,OAAO,KAAK,IAAI,EACxC,CAAC;gBACD,OAAO,CAAC,KAAK,CACX,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,0BAA0B,MAAM,mDAAmD,CAAC,CACpG,CAAC;gBACF,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,iBAAiB,UAAU,CAAC,cAAc,CAAC,EAAE,CAAC,CAAC,CAAC;gBACxE,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,gDAAgD,YAAY,IAAI,CAAC,CAAC,CAAC;gBAC9F,MAAM,IAAI,KAAK,CAAC,2BAA2B,IAAI,6CAA6C,CAAC,CAAC;YAChG,CAAC;YAED,4BAA4B;YAC5B,WAAW,CAAC,YAAY,EAAE,UAAU,CAAC,cAAc,CAAC,CAAC,CAAC;YACtD,OAAO,CAAC,GAAG,CACT,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,aAAa,CAAC,kBAAkB,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,gBAAgB,KAAK,CAAC,IAAI,CAC9F,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,YAAY,CAAC,CAC3C,IAAI,CACN,CAAC;QACJ,CAAC;IACH,CAAC;YAAS,CAAC;QACT,MAAM,OAAO,CAAC,KAAK,EAAE,CAAC;IACxB,CAAC;IAED,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,gDAAgD,CAAC,CAAC,CAAC;AAClF,CAAC"}

package/dist/cli.d.ts

@@ -0,0 +1,3 @@
+#!/usr/bin/env node
+export {};
+//# sourceMappingURL=cli.d.ts.map

package/dist/cli.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"cli.d.ts","sourceRoot":"","sources":["../src/cli.ts"],"names":[],"mappings":""}

package/dist/cli.js

@@ -0,0 +1,103 @@
+#!/usr/bin/env node
+import { Command } from 'commander';
+import chalk from 'chalk';
+import { runTests } from './runner.js';
+import { approveTool } from './approver.js';
+import { initConfig } from './initializer.js';
+const program = new Command();
+program
+    .name('mcpdrift')
+    .description('CI/CD regression and snapshot tester for Model Context Protocol (MCP) servers')
+    .version('1.0.0');
+program
+    .command('run')
+    .description('Run regression tests against the MCP server tools and compare with baseline snapshots')
+    .option('-c, --config <path>', 'path to mcpdrift.config.json')
+    .option('-t, --tool <name>', 'run tests only for a specific tool')
+    .option('--case <id>', 'run tests only for a specific test case ID')
+    .option('--ci', 'strict CI mode: fail if any baseline snapshot is missing, never write baselines')
+    .option('--report <path>', 'output machine-readable JSON report to the specified file path')
+    .option('--json-report <path>', 'alias for --report')
+    .option('--timeout <ms>', 'timeout in milliseconds per tool call', (val) => parseInt(val, 10))
+    .action(async (options) => {
+    try {
+        const summary = await runTests({
+            config: options.config,
+            tool: options.tool,
+            caseId: options.case,
+            ci: options.ci,
+            report: options.report || options.jsonReport,
+            timeout: options.timeout,
+        });
+        if (!summary.success) {
+            process.exit(1);
+        }
+        process.exit(0);
+    }
+    catch (err) {
+        console.error(chalk.red.bold(`\n❌ Error: ${err.message}\n`));
+        process.exit(1);
+    }
+});
+program
+    .command('approve [target]')
+    .description('Overwrite baseline snapshot(s) with the latest server response')
+    .option('-c, --config <path>', 'path to mcpdrift.config.json')
+    .option('--case <id>', 'approve only a specific test case ID')
+    .option('-t, --tool <name>', 'approve all test cases for a specific tool')
+    .option('-a, --all', 'approve all tools configured in mcpdrift.config.json')
+    .action(async (target, options) => {
+    try {
+        await approveTool(target, {
+            config: options.config,
+            caseId: options.case,
+            tool: options.tool,
+            all: options.all,
+        });
+        process.exit(0);
+    }
+    catch (err) {
+        console.error(chalk.red.bold(`\n❌ Approve failed: ${err.message}\n`));
+        process.exit(1);
+    }
+});
+program
+    .command('init')
+    .description('Inspect a running MCP server and generate a starter mcpdrift.config.json')
+    .option('-s, --server <command>', 'server command to start MCP server (default: "node dist/server.js")')
+    .option('--tools <list>', 'comma-separated list of tools to include in config')
+    .option('--all-tools', 'explicitly include all discovered tools in generated test cases')
+    .option('--template', 'generate a starter template config without connecting to a server')
+    .option('-o, --output <path>', 'output config file path (default: "mcpdrift.config.json")')
+    .option('-f, --force', 'overwrite existing config file if present')
+    .action(async (options) => {
+    try {
+        const toolList = options.tools
+            ? options.tools.split(',').map((t) => t.trim()).filter(Boolean)
+            : undefined;
+        await initConfig({
+            server: options.server,
+            tools: toolList,
+            allTools: options.allTools,
+            template: options.template,
+            output: options.output,
+            force: options.force,
+        });
+        process.exit(0);
+    }
+    catch (err) {
+        console.error(chalk.red.bold(`\n❌ Init failed: ${err.message}\n`));
+        process.exit(1);
+    }
+});
+// If no subcommands provided, default to help
+if (process.argv.length <= 2) {
+    program.help();
+}
+else {
+    program.parseAsync(process.argv).catch((err) => {
+        console.error(chalk.red.bold(`Unexpected error: ${err.message}`));
+        process.exit(1);
+    });
+}
+//# sourceMappingURL=cli.js.map