@kryptosai/mcp-observatory 0.23.0 → 0.24.0

package/docs/safety-index/artifacts/sequential-thinking-server.md

@@ -0,0 +1,156 @@
+# MCP Observatory Run Report
+
+Generated at 2026-06-24T02:07:22.515Z
+
+## Target and Environment Metadata
+
+- Target: `sequential-thinking-server`
+- Adapter: `local-process`
+- Command: `npx -y @modelcontextprotocol/server-sequential-thinking`
+- Server: `sequential-thinking-server 0.2.0`
+- Platform: `darwin 25.5.0`
+- Node: `v22.22.1`
+
+## Executive Summary
+
+**Health Score: 93/100 (A)**
+
+| Dimension | Score | Weight |
+| --- | --- | --- |
+| Protocol Compliance | 100/100 | 30% |
+| Schema Quality | 100/100 | 20% |
+| Security | 100/100 | 20% |
+| Reliability | 67/100 | 20% |
+| Performance | 100/100 | 10% |
+
+| Gate | Total | Pass | Fail | Partial | Unsupported | Flaky | Skipped |
+| --- | --- | --- | --- | --- | --- | --- | --- |
+| pass | 7 | 5 | 0 | 0 | 2 | 0 | 0 |
+
+## At a Glance
+
+- Safety verdict: **Needs review** — The server is usable, but caveated checks should be reviewed before agents depend on it.
+- Top risks: No high-priority risks detected.
+- Regression/schema drift: Run `mcp-observatory diff <previous-run.json> <current-run.json>` to classify regressions and schema drift.
+- Failing checks: none
+- Partial or flaky checks: none
+- Skipped checks: none
+- Unsupported checks: prompts, resources
+- Suggested next step: Confirm that unsupported capabilities are intentional for this target: prompts, resources.
+- CI next step: `Add CI: npx @kryptosai/mcp-observatory init-ci --all --command "npx -y <server-package>"`
+
+## Regressions and Recoveries
+
+_Use the `diff` command against another run artifact to classify regressions and recoveries over time._
+
+## Full Capability Status Table
+
+| Focus | Check | Status | Duration (ms) | Message |
+| --- | --- | --- | --- | --- |
+| healthy | conformance | pass | 2.25 | All 7 conformance checks passed. |
+| healthy | schema-quality | pass | 0.89 | All 1 item(s) have good schema quality. |
+| healthy | security | pass | 1.13 | No security issues detected. |
+| healthy | security-lite | pass | 0.11 | No security issues detected (lightweight scan). |
+| healthy | tools | pass | 4.48 | Advertised capability responded with the minimal expected shape (1 item). |
+| confirm intent | prompts | unsupported | 0.01 | Prompts are not advertised by the target. |
+| confirm intent | resources | unsupported | 0.00 | Resources are not advertised by the target. |
+
+## Evidence Snippets
+
+### conformance — pass
+
+Summary: All 7 conformance checks passed.
+
+- Endpoint: `conformance/check`
+  - Advertised: `true`
+  - Responded: `true`
+  - Minimal shape present: `true`
+  - Item count: `7`
+  - Identifiers: none
+  - Diagnostics: [pass] capabilities-present: Server returned capabilities object., [pass] server-info: Server provided initialization info., [pass] tools-capability-match: tools/list returned 1 tool(s). (+4 more)
+
+### schema-quality — pass
+
+Summary: All 1 item(s) have good schema quality.
+
+- Endpoint: `schema-quality/scan`
+  - Advertised: `true`
+  - Responded: `true`
+  - Minimal shape present: `true`
+  - Item count: `0`
+  - Identifiers: none
+  - Diagnostics: none
+
+### security — pass
+
+Summary: No security issues detected.
+
+- Endpoint: `security/scan`
+  - Advertised: `true`
+  - Responded: `true`
+  - Minimal shape present: `true`
+  - Item count: `0`
+  - Identifiers: none
+  - Diagnostics: none
+
+### security-lite — pass
+
+Summary: No security issues detected (lightweight scan).
+
+- Endpoint: `security/scan-lite`
+  - Advertised: `true`
+  - Responded: `true`
+  - Minimal shape present: `true`
+  - Item count: `0`
+  - Identifiers: none
+  - Diagnostics: none
+
+### tools — pass
+
+Summary: Advertised capability responded with the minimal expected shape (1 item).
+
+- Endpoint: `tools/list`
+  - Advertised: `true`
+  - Responded: `true`
+  - Minimal shape present: `true`
+  - Item count: `1`
+  - Identifiers: sequentialthinking
+  - Diagnostics: Sequential Thinking MCP Server running on stdio
+
+### prompts — unsupported
+
+Summary: Prompts are not advertised by the target.
+
+- Endpoint: `prompts/list`
+  - Advertised: `false`
+  - Responded: `false`
+  - Minimal shape present: `false`
+  - Item count: `0`
+  - Identifiers: none
+  - Diagnostics: none
+
+### resources — unsupported
+
+Summary: Resources are not advertised by the target.
+
+- Endpoint: `resources/list | resources/templates/list`
+  - Advertised: `false`
+  - Responded: `false`
+  - Minimal shape present: `false`
+  - Item count: `0`
+  - Identifiers: none
+  - Diagnostics: none
+
+## Reproduction Commands
+
+```bash
+npm run cli -- run --target <path-to-target-config.json>
+npm run cli -- report --run <path-to-run-artifact.json> --format markdown
+```
+
+## Artifact Provenance
+
+- Artifact type: `run`
+- Schema version: `1.0.0`
+- Run ID: `run_2026-06-24T020722515Z_10f2e6a0`
+- Gate: `pass`

package/docs/safety-index/maintainer-note-template.md

@@ -0,0 +1,25 @@
+# Maintainer Note Template
+
+Subject: Reproducible MCP readiness report for `<server>`
+
+Hi,
+
+I ran MCP Observatory against `<server>` as part of the MCP Server Safety Index. This is not a vulnerability report or a drive-by badge request. It is a reproducible compatibility/security-readiness check for MCP servers before agents depend on them.
+
+Report:
+
+- command: `<command>`
+- verdict: `<verdict>`
+- failure class: `<failure-class>`
+- JSON artifact: `<artifact-link>`
+- Markdown report: `<report-link>`
+
+The main thing the report shows is:
+
+> `<one-sentence-finding>`
+
+If useful, I can open a small PR that adds a read-only GitHub Action for this check. If the published package is not the right target, I can instead use the repo's local build/start command so CI validates pull request code.
+
+No account is required. The generated workflow is read-only by default, and strict repos can pin the action to a full commit SHA.
+
+Thanks for maintaining the MCP ecosystem.

package/docs/safety-index/targets.json

@@ -0,0 +1,192 @@
+[
+  {
+    "id": "everything-server",
+    "name": "Official everything server",
+    "repo": "https://github.com/modelcontextprotocol/servers",
+    "packageName": "@modelcontextprotocol/server-everything",
+    "category": "Reference",
+    "command": "npx",
+    "args": ["-y", "@modelcontextprotocol/server-everything"],
+    "timeoutMs": 60000,
+    "riskClass": "Reference compatibility",
+    "failureClass": "Broad protocol surface",
+    "whyItMatters": "Exercises tools, prompts, and resources in one official reference server.",
+    "reproductionNotes": "Zero-config official package; useful as a broad protocol baseline.",
+    "publicProof": "https://github.com/modelcontextprotocol/servers/pull/4392"
+  },
+  {
+    "id": "sequential-thinking-server",
+    "name": "Official sequential thinking server",
+    "repo": "https://github.com/modelcontextprotocol/servers",
+    "packageName": "@modelcontextprotocol/server-sequential-thinking",
+    "category": "Reference",
+    "command": "npx",
+    "args": ["-y", "@modelcontextprotocol/server-sequential-thinking"],
+    "timeoutMs": 60000,
+    "riskClass": "Reference compatibility",
+    "failureClass": "Tool schema clarity",
+    "whyItMatters": "Small official tools server that makes schema readability easy to inspect.",
+    "reproductionNotes": "Zero-config official package."
+  },
+  {
+    "id": "memory-server",
+    "name": "Official memory server",
+    "repo": "https://github.com/modelcontextprotocol/servers",
+    "packageName": "@modelcontextprotocol/server-memory",
+    "category": "Reference / Memory",
+    "command": "npx",
+    "args": ["-y", "@modelcontextprotocol/server-memory"],
+    "timeoutMs": 60000,
+    "riskClass": "Agent memory mutation",
+    "failureClass": "Persistent state tools",
+    "whyItMatters": "Memory tools are agent-facing state; their schemas and write behavior should be visible before use.",
+    "reproductionNotes": "Zero-config official package."
+  },
+  {
+    "id": "filesystem-server",
+    "name": "Official filesystem server",
+    "repo": "https://github.com/modelcontextprotocol/servers",
+    "packageName": "@modelcontextprotocol/server-filesystem",
+    "category": "Filesystem",
+    "command": "npx",
+    "args": ["-y", "@modelcontextprotocol/server-filesystem", "examples/filesystem-fixture"],
+    "timeoutMs": 30000,
+    "riskClass": "Filesystem boundary",
+    "failureClass": "Sandboxed filesystem access",
+    "whyItMatters": "Filesystem tools need harmless test roots and clear read/write boundaries.",
+    "reproductionNotes": "Runs against the checked-in harmless fixture directory."
+  },
+  {
+    "id": "context7-server",
+    "name": "Context7",
+    "repo": "https://github.com/upstash/context7",
+    "packageName": "@upstash/context7-mcp",
+    "category": "Documentation / Search",
+    "command": "npx",
+    "args": ["-y", "@upstash/context7-mcp"],
+    "timeoutMs": 60000,
+    "riskClass": "Untrusted content retrieval",
+    "failureClass": "Prompt-injection-sensitive retrieval",
+    "whyItMatters": "Documentation retrieval tools can return untrusted text into agent context.",
+    "reproductionNotes": "Zero-config public package."
+  },
+  {
+    "id": "promptopia-server",
+    "name": "Promptopia",
+    "repo": "https://www.npmjs.com/package/promptopia-mcp",
+    "packageName": "promptopia-mcp",
+    "category": "Prompts",
+    "command": "npx",
+    "args": ["-y", "promptopia-mcp"],
+    "env": {
+      "PROMPTS_DIR": "examples/promptopia-prompts"
+    },
+    "timeoutMs": 30000,
+    "riskClass": "Prompt surface",
+    "failureClass": "Prompt/resource contract",
+    "whyItMatters": "Prompt-serving MCP packages need predictable prompt inventory and startup behavior.",
+    "reproductionNotes": "Uses the checked-in prompt fixture through package defaults."
+  },
+  {
+    "id": "ref-tools-server",
+    "name": "Ref tools",
+    "repo": "https://www.npmjs.com/package/ref-tools-mcp",
+    "packageName": "ref-tools-mcp",
+    "category": "Developer Tools",
+    "command": "npx",
+    "args": ["-y", "ref-tools-mcp"],
+    "timeoutMs": 60000,
+    "riskClass": "Developer reference tooling",
+    "failureClass": "Prompt/tool inventory",
+    "whyItMatters": "Developer tools become agent dependencies when wired into coding workflows.",
+    "reproductionNotes": "Zero-config public package."
+  },
+  {
+    "id": "opentofu-server",
+    "name": "OpenTofu MCP server",
+    "repo": "https://github.com/opentofu/opentofu-mcp-server",
+    "packageName": "@opentofu/opentofu-mcp-server",
+    "category": "Infrastructure",
+    "command": "npx",
+    "args": ["-y", "@opentofu/opentofu-mcp-server"],
+    "timeoutMs": 60000,
+    "riskClass": "Infrastructure-as-code",
+    "failureClass": "Infrastructure tool surface",
+    "whyItMatters": "Infrastructure tools can influence production configuration and should have reviewable MCP surfaces.",
+    "reproductionNotes": "Zero-config public package."
+  },
+  {
+    "id": "puppeteer-server",
+    "name": "Puppeteer MCP server",
+    "repo": "https://www.npmjs.com/package/puppeteer-mcp-server",
+    "packageName": "puppeteer-mcp-server",
+    "category": "Browser Automation",
+    "command": "npx",
+    "args": ["-y", "puppeteer-mcp-server"],
+    "timeoutMs": 60000,
+    "securitySuppressions": ["puppeteer_evaluate:shell-injection"],
+    "riskClass": "Browser control",
+    "failureClass": "Browser/code execution boundary",
+    "whyItMatters": "Browser automation gives agents navigation, page state, and sometimes code-evaluation powers.",
+    "reproductionNotes": "Intentional browser evaluation is suppressed so remaining findings stay readable."
+  },
+  {
+    "id": "browsermcp-server",
+    "name": "BrowserMCP",
+    "repo": "https://github.com/BrowserMCP/mcp",
+    "packageName": "@browsermcp/mcp",
+    "category": "Browser Automation",
+    "command": "npx",
+    "args": ["-y", "@browsermcp/mcp"],
+    "timeoutMs": 60000,
+    "riskClass": "Browser control",
+    "failureClass": "Browser-control boundary",
+    "whyItMatters": "Browser-control servers need visible tool boundaries before agents drive real browsing sessions.",
+    "reproductionNotes": "Zero-config public package.",
+    "publicProof": "https://github.com/BrowserMCP/mcp/pull/189"
+  },
+  {
+    "id": "playwright-mcp-server",
+    "name": "Microsoft Playwright MCP",
+    "repo": "https://github.com/microsoft/playwright-mcp",
+    "packageName": "@playwright/mcp",
+    "category": "Browser Automation",
+    "command": "npx",
+    "args": ["-y", "@playwright/mcp"],
+    "timeoutMs": 60000,
+    "riskClass": "Browser control",
+    "failureClass": "Browser/code execution boundary",
+    "whyItMatters": "Popular browser automation servers need explicit review around navigation, screenshots, and code evaluation.",
+    "reproductionNotes": "Zero-config public package; security findings represent policy-review prompts, not a vulnerability claim."
+  },
+  {
+    "id": "antv-chart-server",
+    "name": "AntV chart MCP server",
+    "repo": "https://github.com/antvis/mcp-server-chart",
+    "packageName": "@antv/mcp-server-chart",
+    "category": "Visualization",
+    "command": "npx",
+    "args": ["-y", "@antv/mcp-server-chart"],
+    "timeoutMs": 60000,
+    "riskClass": "Generated artifacts",
+    "failureClass": "Artifact-producing tools",
+    "whyItMatters": "Visualization servers produce files or structured outputs agents may pass into workflows.",
+    "reproductionNotes": "Zero-config public package.",
+    "publicProof": "https://github.com/antvis/mcp-server-chart/pull/312"
+  },
+  {
+    "id": "executeautomation-playwright-server",
+    "name": "ExecuteAutomation Playwright MCP",
+    "repo": "https://github.com/executeautomation/mcp-playwright",
+    "packageName": "@executeautomation/playwright-mcp-server",
+    "category": "Browser Automation",
+    "command": "npx",
+    "args": ["-y", "@executeautomation/playwright-mcp-server"],
+    "timeoutMs": 60000,
+    "riskClass": "Browser control",
+    "failureClass": "Startup/listing reproducibility",
+    "whyItMatters": "A public browser automation package that fails basic listing checks shows why reproducible readiness evidence matters.",
+    "reproductionNotes": "Evaluated as a public package; current result should be treated as a maintainer conversation starter.",
+    "publicProof": "https://github.com/executeautomation/mcp-playwright/pull/225"
+  }
+]

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kryptosai/mcp-observatory",
-  "version": "0.23.0",
+  "version": "0.24.0",
   "description": "The CI and security gate for MCP servers before agents depend on them.",
   "mcpName": "io.github.KryptosAI/mcp-observatory",
   "license": "MIT",
@@ -42,7 +42,13 @@
     "build": "tsc -p tsconfig.json",
     "cli": "tsx src/cli.ts",
     "dashboard:build": "tsx scripts/build-dashboard.ts",
+    "metrics:collect": "tsx scripts/metrics-dashboard.ts collect",
+    "metrics:build": "tsx scripts/metrics-dashboard.ts build",
+    "metrics:refresh": "tsx scripts/metrics-dashboard.ts refresh",
+    "metrics:open": "tsx scripts/metrics-dashboard.ts open",
+    "metrics:serve": "tsx scripts/metrics-dashboard.ts serve",
     "integration:real": "tsx scripts/run-real-server-matrix.ts",
+    "safety-index": "tsx scripts/run-safety-index.ts",
     "lint": "eslint .",
     "prepack": "npm run build",
     "proof:refresh": "tsx scripts/refresh-proof-artifacts.ts",
@@ -95,18 +101,15 @@
   },
   "devDependencies": {
     "@eslint/js": "10.0.1",
-    "@semantic-release/changelog": "^6.0.3",
-    "@semantic-release/git": "^10.0.1",
-    "@types/node": "25.9.3",
-    "@typescript-eslint/eslint-plugin": "8.61.1",
-    "@typescript-eslint/parser": "8.61.1",
+    "@types/node": "26.0.0",
+    "@typescript-eslint/eslint-plugin": "8.62.0",
+    "@typescript-eslint/parser": "8.62.0",
     "ajv": "8.20.0",
     "eslint": "10.5.0",
-    "globals": "17.6.0",
-    "semantic-release": "^25.0.5",
+    "globals": "17.7.0",
     "tsx": "4.22.4",
     "typescript": "6.0.3",
-    "typescript-eslint": "8.61.1",
+    "typescript-eslint": "8.62.0",
     "vitest": "4.1.9"
   }
 }