@kryptosai/mcp-observatory 1.32.0 → 1.33.0

package/docs/safety-index/artifacts/kubernetes-server.md

@@ -1,13 +1,13 @@
 # MCP Observatory Run Report
 
-Generated at 2026-07-06T19:49:58.878Z
+Generated at 2026-07-15T22:33:34.012Z
 
 ## Target and Environment Metadata
 
 - Target: `kubernetes-server`
 - Adapter: `local-process`
 - Command: `npx -y mcp-server-kubernetes`
-- Server: `kubernetes 3.9.3`
+- Server: `kubernetes 4.0.4`
 - Platform: `darwin 25.5.0`
 - Node: `v22.22.1`
 
@@ -25,20 +25,130 @@ Generated at 2026-07-06T19:49:58.878Z
 
 | Gate | Total | Pass | Fail | Partial | Unsupported | Flaky | Skipped |
 | --- | --- | --- | --- | --- | --- | --- | --- |
-| fail | 8 | 4 | 3 | 1 | 0 | 0 | 0 |
+| fail | 9 | 4 | 3 | 2 | 0 | 0 | 0 |
 
 ## At a Glance
 
 - Safety verdict: **Blocked** — One or more checks can break agent dependence and should be fixed before production use.
-- Top risks: schema-quality: Found 7 quality finding(s) across 29 item(s): 0 warnings, 7 info.; attack-sim: Safe attack simulation found 5 finding(s): 3 high, 2 medium, 0 low.; security: Found 8 security finding(s): 3 high, 3 medium, 2 low.
+- Top risks: runtime-profile: Detected 11 potential egress target(s) and 76 potential state mutation(s) with high confidence.; schema-quality: Found 7 quality finding(s) across 29 item(s): 0 warnings, 7 info.; attack-sim: Safe attack simulation found 5 finding(s): 3 high, 2 medium, 0 low.
 - Regression/schema drift: Run `mcp-observatory diff <previous-run.json> <current-run.json>` to classify regressions and schema drift.
 - Failing checks: security-lite, security, attack-sim
-- Partial or flaky checks: schema-quality
+- Partial or flaky checks: runtime-profile, schema-quality
 - Skipped checks: none
 - Unsupported checks: none
 - Suggested next step: Start with the failing checks: security-lite, security, attack-sim.
 - CI next step: `Add CI: npx @kryptosai/mcp-observatory setup-ci --all --command "npx -y <server-package>"`
 
+## What Was Not Tested
+
+- ℹ️ credential_access: Credential scanning was performed (see security findings)
+- ℹ️ destructive_payloads: Destructive payloads were not attempted (safe-mode only)
+
+## Runtime Profile
+
+### Egress Manifest
+
+The following targets were identified as potentially reachable by this server (confidence: **high**):
+
+| Target | Protocol | Source | Confidence |
+| --- | --- | --- | --- |
+| If true, immediately remove resources from API and bypass graceful deletion | unknown | description_analysis | medium |
+| If true, immediately remove resources from API and bypass graceful deletion | unknown | description_analysis | medium |
+| Path to a YAML file to create resources from. The path is read on the machine running the MCP server, so it is rejected when the server runs over a remote (SSE/Streamable HTTP) transport; use 'manifest' to pass the file's contents instead. | unknown | description_analysis | medium |
+| Path to file for creating configmap/secret (e.g. ["key1=/path/to/file1", "key2=/path/to/file2"]). The path is read on the machine running the MCP server, so it is rejected when the server runs over a remote (SSE/Streamable HTTP) transport; use "fromFileContent" to pass file contents directly instead. | unknown | description_analysis | medium |
+| Inline file contents for creating a configmap/secret, provided by the client instead of a server-side path (e.g. [{"key": "app.conf", "content": "..."}]). Safe on all transports; use this instead of "fromFile" on remote (SSE/Streamable HTTP) servers. | unknown | description_analysis | medium |
+| kubectl_reconnect | unknown | description_analysis | low |
+| API version to use (e.g. 'apps/v1') | unknown | description_analysis | medium |
+| Helm repository URL (optional if using local chart path) | unknown | description_analysis | medium |
+| Helm repository URL (optional if using local chart path) | unknown | description_analysis | medium |
+| API group to filter by | unknown | description_analysis | medium |
+| list_api_resources | unknown | description_analysis | low |
+
+### State Mutations
+
+The following state-modifying operations were identified from tool schemas:
+
+| Resource | Operation | Scope | Source |
+| --- | --- | --- | --- |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | tool_schema |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | tool_schema |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | tool_schema |
+| network | write | specific_path | tool_schema |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| network | execute | specific_path | tool_schema |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | execute | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| network | execute | specific_path | tool_schema |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+| filesystem | write | working_directory | description_analysis |
+
+_Analyzed at 2026-07-15T22:33:35.351Z_
+
 ## Regressions and Recoveries
 
 _Use the `diff` command against another run artifact to classify regressions and recoveries over time._
@@ -47,14 +157,15 @@ _Use the `diff` command against another run artifact to classify regressions and
 
 | Focus | Check | Status | Duration (ms) | Message |
 | --- | --- | --- | --- | --- |
-| healthy | conformance | pass | 1.69 | All 7 conformance checks passed. |
-| healthy | prompts | pass | 0.30 | Advertised capability responded with the minimal expected shape (1 item). |
-| healthy | resources | pass | 0.38 | Advertised capability responded with the minimal expected shape, but one optional resource endpoint appears unsupported. |
-| healthy | tools | pass | 1.26 | Advertised capability responded with the minimal expected shape (23 items). |
-| review | schema-quality | partial | 0.49 | Found 7 quality finding(s) across 29 item(s): 0 warnings, 7 info. |
-| act now | attack-sim | fail | 0.75 | Safe attack simulation found 5 finding(s): 3 high, 2 medium, 0 low. |
-| act now | security | fail | 0.28 | Found 8 security finding(s): 3 high, 3 medium, 2 low. |
-| act now | security-lite | fail | 0.06 | Found 8 security finding(s): 3 high, 3 medium, 2 low. |
+| healthy | conformance | pass | 1.79 | All 7 conformance checks passed. |
+| healthy | prompts | pass | 0.31 | Advertised capability responded with the minimal expected shape (1 item). |
+| healthy | resources | pass | 0.39 | Advertised capability responded with the minimal expected shape, but one optional resource endpoint appears unsupported. |
+| healthy | tools | pass | 1.27 | Advertised capability responded with the minimal expected shape (23 items). |
+| review | runtime-profile | partial | 0.30 | Detected 11 potential egress target(s) and 76 potential state mutation(s) with high confidence. |
+| review | schema-quality | partial | 0.97 | Found 7 quality finding(s) across 29 item(s): 0 warnings, 7 info. |
+| act now | attack-sim | fail | 0.96 | Safe attack simulation found 5 finding(s): 3 high, 2 medium, 0 low. |
+| act now | security | fail | 0.35 | Found 8 security finding(s): 3 high, 3 medium, 2 low. |
+| act now | security-lite | fail | 0.07 | Found 8 security finding(s): 3 high, 3 medium, 2 low. |
 
 ## Evidence Snippets
 
@@ -80,7 +191,7 @@ Summary: Advertised capability responded with the minimal expected shape (1 item
   - Minimal shape present: `true`
   - Item count: `1`
   - Identifiers: k8s-diagnose
-  - Diagnostics: Starting Kubernetes MCP server v3.9.3, handling commands..., Telemetry: Disabled, (node:13988) [DEP0040] DeprecationWarning: The `punycode` module is deprecated. Please use a userland alternative instead. (+1 more)
+  - Diagnostics: Starting Kubernetes MCP server v4.0.4, handling commands..., Telemetry: Disabled, (node:78269) [DEP0040] DeprecationWarning: The `punycode` module is deprecated. Please use a userland alternative instead. (+1 more)
 
 ### resources — pass
 
@@ -92,14 +203,14 @@ Summary: Advertised capability responded with the minimal expected shape, but on
   - Minimal shape present: `true`
   - Item count: `5`
   - Identifiers: k8s://default/pods, k8s://default/deployments, k8s://default/services, k8s://namespaces, k8s://nodes
-  - Diagnostics: Starting Kubernetes MCP server v3.9.3, handling commands..., Telemetry: Disabled, (node:13988) [DEP0040] DeprecationWarning: The `punycode` module is deprecated. Please use a userland alternative instead. (+1 more)
+  - Diagnostics: Starting Kubernetes MCP server v4.0.4, handling commands..., Telemetry: Disabled, (node:78269) [DEP0040] DeprecationWarning: The `punycode` module is deprecated. Please use a userland alternative instead. (+1 more)
 - Endpoint: `resources/templates/list`
   - Advertised: `true`
   - Responded: `false`
   - Minimal shape present: `false`
   - Item count: `0`
   - Identifiers: none
-  - Diagnostics: MCP error -32601: Method not found, Starting Kubernetes MCP server v3.9.3, handling commands..., Telemetry: Disabled (+2 more)
+  - Diagnostics: MCP error -32601: Method not found, Starting Kubernetes MCP server v4.0.4, handling commands..., Telemetry: Disabled (+2 more)
 
 ### tools — pass
 
@@ -111,7 +222,19 @@ Summary: Advertised capability responded with the minimal expected shape (23 ite
   - Minimal shape present: `true`
   - Item count: `23`
   - Identifiers: cleanup, kubectl_get, kubectl_describe, kubectl_apply, kubectl_delete (+18 more)
-  - Diagnostics: Starting Kubernetes MCP server v3.9.3, handling commands..., Telemetry: Disabled, (node:13988) [DEP0040] DeprecationWarning: The `punycode` module is deprecated. Please use a userland alternative instead. (+1 more)
+  - Diagnostics: Starting Kubernetes MCP server v4.0.4, handling commands..., Telemetry: Disabled, (node:78269) [DEP0040] DeprecationWarning: The `punycode` module is deprecated. Please use a userland alternative instead. (+1 more)
+
+### runtime-profile — partial
+
+Summary: Detected 11 potential egress target(s) and 76 potential state mutation(s) with high confidence.
+
+- Endpoint: `runtime-profile/analyze`
+  - Advertised: `true`
+  - Responded: `true`
+  - Minimal shape present: `true`
+  - Item count: `87`
+  - Identifiers: none
+  - Diagnostics: Egress entries: 11, State mutations: 76, Confidence: high
 
 ### schema-quality — partial
 
@@ -172,5 +295,5 @@ npm run cli -- report --run <path-to-run-artifact.json> --format markdown
 
 - Artifact type: `run`
 - Schema version: `1.0.0`
-- Run ID: `run_2026-07-06T194958878Z_b41bb221`
+- Run ID: `run_2026-07-15T223334012Z_6b45db48`
 - Gate: `fail`

package/docs/safety-index/artifacts/langfuse-mcp.json

@@ -0,0 +1,66 @@
+{
+  "artifactType": "run",
+  "schemaVersion": "1.0.0",
+  "gate": "fail",
+  "runId": "run_2026-07-15T223456283Z_8894143f",
+  "createdAt": "2026-07-15T22:34:56.283Z",
+  "toolVersion": "1.32.1",
+  "target": {
+    "targetId": "langfuse-mcp",
+    "adapter": "local-process",
+    "command": "npx",
+    "args": [
+      "-y",
+      "langfuse-mcp"
+    ],
+    "cwd": ".",
+    "metadata": {
+      "package": "langfuse-mcp",
+      "purpose": "mcp-safety-index",
+      "riskClass": "LLM observability access",
+      "failureClass": "Trace/observation data boundary",
+      "whyItMatters": "Langfuse MCP servers can access LLM traces, prompt versions, and evaluation data — agents could read sensitive prompt/output data from production LLM pipelines."
+    }
+  },
+  "environment": {
+    "platform": "darwin 25.5.0",
+    "nodeVersion": "v22.22.1"
+  },
+  "summary": {
+    "total": 3,
+    "pass": 0,
+    "fail": 0,
+    "partial": 0,
+    "unsupported": 0,
+    "flaky": 0,
+    "skipped": 3,
+    "gate": "fail"
+  },
+  "checks": [
+    {
+      "id": "tools",
+      "capability": "tools",
+      "status": "skipped",
+      "durationMs": 0,
+      "message": "Skipped because startup failed before the MCP session initialized. See the failure diagnosis.",
+      "evidence": []
+    },
+    {
+      "id": "prompts",
+      "capability": "prompts",
+      "status": "skipped",
+      "durationMs": 0,
+      "message": "Skipped because startup failed before the MCP session initialized. See the failure diagnosis.",
+      "evidence": []
+    },
+    {
+      "id": "resources",
+      "capability": "resources",
+      "status": "skipped",
+      "durationMs": 0,
+      "message": "Skipped because startup failed before the MCP session initialized. See the failure diagnosis.",
+      "evidence": []
+    }
+  ],
+  "fatalError": "Could not establish a plain stdio MCP session for target `langfuse-mcp`.\nCommand: npx -y langfuse-mcp\nDiagnosis: The process exited or detached before MCP initialization completed.\nRaw error: MCP error -32000: Connection closed\nLikely causes:\n- The package may expect extra startup arguments or environment variables.\n- The package may not behave like a plain local-process stdio target under this invocation.\n- The server may be crashing immediately and only leaving clues on stderr.\nNext steps:\n- Run the command manually and look for usage output, auth prompts, or crash text.\n- Check whether the package expects a different transport or an app-oriented startup flow.\n- Use the recent stderr lines below before assuming this is a harness bug.\nRecent stderr:\n- langfuse-mcp: Invalid Langfuse configuration:\n- - LANGFUSE_PUBLIC_KEY: Required\n- - LANGFUSE_SECRET_KEY: Required\n- - LANGFUSE_BASE_URL: Required\n- See .env.example for the expected variables."
+}

package/docs/safety-index/artifacts/langfuse-mcp.md

@@ -0,0 +1,106 @@
+# MCP Observatory Run Report
+
+Generated at 2026-07-15T22:34:56.283Z
+
+## Target and Environment Metadata
+
+- Target: `langfuse-mcp`
+- Adapter: `local-process`
+- Command: `npx -y langfuse-mcp`
+- Server: `unknown `
+- Platform: `darwin 25.5.0`
+- Node: `v22.22.1`
+
+## Executive Summary
+
+| Gate | Total | Pass | Fail | Partial | Unsupported | Flaky | Skipped |
+| --- | --- | --- | --- | --- | --- | --- | --- |
+| fail | 3 | 0 | 0 | 0 | 0 | 0 | 3 |
+
+## At a Glance
+
+- Safety verdict: **Blocked** — One or more checks can break agent dependence and should be fixed before production use.
+- Top risks: startup: server failed to start
+- 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: tools, prompts, resources
+- Unsupported checks: none
+- Suggested next step: Run the target command manually, compare stderr with the diagnosis below, and only raise timeoutMs if startup is genuinely slow.
+- CI next step: `Add CI: npx @kryptosai/mcp-observatory setup-ci --all --command "npx -y <server-package>"`
+
+## What Was Not Tested
+
+- 🔒 network_egress: No outbound network calls were attempted during scan
+- 🔒 filesystem_mutation: Filesystem write operations were not exercised
+- 🔒 credential_access: Credential scanning was not performed
+- ℹ️ destructive_payloads: Destructive payloads were not attempted (safe-mode only)
+
+## Regressions and Recoveries
+
+_Use the `diff` command against another run artifact to classify regressions and recoveries over time._
+
+## Failure Diagnosis
+
+```text
+Could not establish a plain stdio MCP session for target `langfuse-mcp`.
+Command: npx -y langfuse-mcp
+Diagnosis: The process exited or detached before MCP initialization completed.
+Raw error: MCP error -32000: Connection closed
+Likely causes:
+- The package may expect extra startup arguments or environment variables.
+- The package may not behave like a plain local-process stdio target under this invocation.
+- The server may be crashing immediately and only leaving clues on stderr.
+Next steps:
+- Run the command manually and look for usage output, auth prompts, or crash text.
+- Check whether the package expects a different transport or an app-oriented startup flow.
+- Use the recent stderr lines below before assuming this is a harness bug.
+Recent stderr:
+- langfuse-mcp: Invalid Langfuse configuration:
+- - LANGFUSE_PUBLIC_KEY: Required
+- - LANGFUSE_SECRET_KEY: Required
+- - LANGFUSE_BASE_URL: Required
+- See .env.example for the expected variables.
+```
+
+## Full Capability Status Table
+
+| Focus | Check | Status | Duration (ms) | Message |
+| --- | --- | --- | --- | --- |
+| inspect startup | prompts | skipped | 0.00 | Skipped because startup failed before the MCP session initialized. See the failure diagnosis. |
+| inspect startup | resources | skipped | 0.00 | Skipped because startup failed before the MCP session initialized. See the failure diagnosis. |
+| inspect startup | tools | skipped | 0.00 | Skipped because startup failed before the MCP session initialized. See the failure diagnosis. |
+
+## Evidence Snippets
+
+### prompts — skipped
+
+Summary: Skipped because startup failed before the MCP session initialized. See the failure diagnosis.
+
+_No evidence was captured._
+
+### resources — skipped
+
+Summary: Skipped because startup failed before the MCP session initialized. See the failure diagnosis.
+
+_No evidence was captured._
+
+### tools — skipped
+
+Summary: Skipped because startup failed before the MCP session initialized. See the failure diagnosis.
+
+_No evidence was captured._
+
+## 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_2026-07-15T223456283Z_8894143f`
+- Gate: `fail`