Improve LLM tool parameter guidance and add E2E testing framework #26
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -57,6 +57,10 @@ helm-lint: ## Run helm lint for Helm chart | |
| test: ## Run unit tests | ||
| $(GOTEST) -v ./... | ||
|
|
||
| .PHONY: e2e-test | ||
| e2e-test: ## Run E2E tests | ||
| @cd e2e-tests && ./scripts/run-tests.sh | ||
|
|
||
| .PHONY: test-coverage-and-junit | ||
| test-coverage-and-junit: ## Run unit tests with coverage and junit output | ||
| go install github.com/jstemmer/go-junit-report/[email protected] | ||
|
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,92 @@ | ||
| # StackRox MCP E2E Testing | ||
|
|
||
| End-to-end tests for the StackRox MCP server using [gevals](https://github.com/genmcp/gevals). | ||
|
|
||
| ## Prerequisites | ||
|
|
||
| - Go 1.25+ | ||
| - OpenAI API Key (for AI agent and LLM judge) | ||
| - StackRox API Token | ||
|
|
||
| ## Setup | ||
|
|
||
| ### 1. Build gevals | ||
|
|
||
| ```bash | ||
| cd e2e-tests | ||
| ./scripts/build-gevals.sh | ||
| ``` | ||
|
|
||
| ### 2. Configure Environment | ||
|
|
||
| Create `.env` file: | ||
|
|
||
| ```bash | ||
| OPENAI_API_KEY=<OpenAI Key> | ||
| STACKROX_MCP__CENTRAL__API_TOKEN=<StackRox API Token> | ||
| ``` | ||
|
|
||
| ## Running Tests | ||
|
|
||
| ```bash | ||
| ./scripts/run-tests.sh | ||
| ``` | ||
|
|
||
| Results are saved to `gevals-stackrox-mcp-e2e-out.json`. | ||
|
|
||
| ### View Results | ||
|
|
||
| ```bash | ||
| # Summary | ||
| jq '.tasks[] | {name, passed}' gevals-stackrox-mcp-e2e-out.json | ||
|
|
||
| # Tool calls | ||
| jq '.tasks[].callHistory[] | {toolName, arguments}' gevals-stackrox-mcp-e2e-out.json | ||
|
Comment on lines
+41
to
+44
Collaborator
There was a problem hiding this comment.
(we are in
(there is not JSON property
(keys are
I didn't test |
||
| ``` | ||
|
|
||
| ## Test Cases | ||
|
|
||
| | Test | Description | Tool | | ||
| |------|-------------|------| | ||
| | `list-clusters` | List all clusters | `list_clusters` | | ||
| | `cve-detected-workloads` | CVE detected in deployments | `get_deployments_for_cve` | | ||
| | `cve-detected-clusters` | CVE detected in clusters | `get_clusters_with_orchestrator_cve` | | ||
| | `cve-nonexistent` | Handle non-existent CVE | `get_clusters_with_orchestrator_cve` | | ||
| | `cve-cluster-does-exist` | CVE with cluster filter | `get_clusters_with_orchestrator_cve` | | ||
| | `cve-cluster-does-not-exist` | CVE with cluster filter | `get_clusters_with_orchestrator_cve` | | ||
| | `cve-clusters-general` | General CVE query | `get_clusters_with_orchestrator_cve` | | ||
| | `cve-cluster-list` | CVE across clusters | `get_clusters_with_orchestrator_cve` | | ||
|
|
||
| ## Configuration | ||
|
|
||
| - **`gevals/eval.yaml`**: Main test configuration, agent settings, assertions | ||
| - **`gevals/mcp-config.yaml`**: MCP server configuration | ||
| - **`gevals/tasks/*.yaml`**: Individual test task definitions | ||
|
|
||
| ## How It Works | ||
|
|
||
| Gevals uses a proxy architecture to intercept MCP tool calls: | ||
|
|
||
| 1. AI agent receives task prompt | ||
| 2. Agent calls MCP tool | ||
| 3. Gevals proxy intercepts and records the call | ||
| 4. Call forwarded to StackRox MCP server | ||
| 5. Server executes and returns result | ||
| 6. Gevals validates assertions and response quality | ||
|
|
||
| ## Troubleshooting | ||
|
|
||
| **Tests fail - no tools called** | ||
| - Verify StackRox Central is accessible | ||
| - Check API token permissions | ||
|
|
||
| **Build errors** | ||
| ```bash | ||
| go mod tidy | ||
|
Collaborator
There was a problem hiding this comment. Can you please check this. For me to direct dependencies. I'm on |
||
| ./scripts/build-gevals.sh | ||
| ``` | ||
|
|
||
| ## Further Reading | ||
|
|
||
| - [Gevals Documentation](https://github.com/genmcp/gevals) | ||
| - [StackRox MCP Server](../README.md) | ||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change | ||||
|---|---|---|---|---|---|---|
| @@ -0,0 +1,101 @@ | ||||||
| kind: Eval | ||||||
| metadata: | ||||||
| name: "stackrox-mcp-e2e" | ||||||
| config: | ||||||
| agent: | ||||||
| type: "builtin.openai-agent" | ||||||
| model: "gpt-4o" | ||||||
| llmJudge: | ||||||
| env: | ||||||
| baseUrlKey: JUDGE_BASE_URL | ||||||
| apiKeyKey: JUDGE_API_KEY | ||||||
| modelNameKey: JUDGE_MODEL_NAME | ||||||
| mcpConfigFile: mcp-config.yaml | ||||||
| taskSets: | ||||||
| # Test 1: List clusters | ||||||
| - path: tasks/list-clusters.yaml | ||||||
| assertions: | ||||||
| toolsUsed: | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "list_clusters" | ||||||
| minToolCalls: 1 | ||||||
| maxToolCalls: 1 | ||||||
|
|
||||||
| # Test 2: CVE detected in workloads | ||||||
| - path: tasks/cve-detected-workloads.yaml | ||||||
| assertions: | ||||||
|
Collaborator
There was a problem hiding this comment. Is it possible to define |
||||||
| toolsUsed: | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "get_deployments_for_cve" | ||||||
| argumentsMatch: | ||||||
| cveName: "CVE-2021-31805" | ||||||
| minToolCalls: 1 | ||||||
| maxToolCalls: 1 | ||||||
|
|
||||||
| # Test 3: CVE detected in clusters - basic | ||||||
| - path: tasks/cve-detected-clusters.yaml | ||||||
| assertions: | ||||||
| toolsUsed: | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "get_clusters_with_orchestrator_cve" | ||||||
| argumentsMatch: | ||||||
| cveName: "CVE-2016-1000031" | ||||||
| minToolCalls: 1 | ||||||
| maxToolCalls: 3 | ||||||
|
Collaborator
There was a problem hiding this comment. What |
||||||
|
|
||||||
| # Test 4: Non-existent CVE | ||||||
| # Expects 3 calls because "Is CVE detected in my clusters?" triggers comprehensive check | ||||||
| # (orchestrator, deployments, nodes). The LLM cannot know beforehand if CVE exists. | ||||||
| - path: tasks/cve-nonexistent.yaml | ||||||
| assertions: | ||||||
| toolsUsed: | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "get_clusters_with_orchestrator_cve" | ||||||
| argumentsMatch: | ||||||
| cveName: "CVE-2099-00001" | ||||||
| minToolCalls: 1 | ||||||
| maxToolCalls: 3 | ||||||
|
|
||||||
| # Test 5: CVE with specific cluster filter (does exist) | ||||||
| - path: tasks/cve-cluster-does-exist.yaml | ||||||
| assertions: | ||||||
| toolsUsed: | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "list_clusters" | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "get_clusters_with_orchestrator_cve" | ||||||
| argumentsMatch: | ||||||
| cveName: "CVE-2016-1000031" | ||||||
| minToolCalls: 1 | ||||||
| maxToolCalls: 2 | ||||||
|
|
||||||
| # Test 6: CVE with specific cluster filter (does not exist) | ||||||
| - path: tasks/cve-cluster-does-not-exist.yaml | ||||||
| assertions: | ||||||
| toolsUsed: | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "list_clusters" | ||||||
| minToolCalls: 1 | ||||||
| maxToolCalls: 2 | ||||||
|
Collaborator
There was a problem hiding this comment. LLM should not fetch list of clusters twice:
Suggested change
|
||||||
|
|
||||||
| # Test 7: CVE detected in clusters - general | ||||||
| - path: tasks/cve-clusters-general.yaml | ||||||
| assertions: | ||||||
| toolsUsed: | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "get_clusters_with_orchestrator_cve" | ||||||
| argumentsMatch: | ||||||
| cveName: "CVE-2021-31805" | ||||||
| minToolCalls: 1 | ||||||
| maxToolCalls: 5 | ||||||
|
|
||||||
| # Test 8: CVE check with cluster list reference | ||||||
| - path: tasks/cve-cluster-list.yaml | ||||||
| assertions: | ||||||
| toolsUsed: | ||||||
| - server: stackrox-mcp | ||||||
| toolPattern: "get_clusters_with_orchestrator_cve" | ||||||
| argumentsMatch: | ||||||
| cveName: "CVE-2024-52577" | ||||||
| minToolCalls: 1 | ||||||
| maxToolCalls: 5 | ||||||
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As I see, this script includes also building of the
gevalsif it's not already built. In short we can set environment variables and call this script.Can we add
maketarget that will rune2e-tests? (more or loss just calls this script, as I understand)