test-genie-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OPENAI_API_KEY | No | OpenAI API key used when provider is openai. | |
| ANTHROPIC_API_KEY | No | Anthropic API key used when provider is anthropic. | |
| TEST_GENIE_STORAGE_DIR | No | Where scenarios / results / iteration logs live. | ~/.test-genie-mcp |
| TEST_GENIE_ALLOWED_ROOT | No | Capability-based path safety — server refuses to read/write outside this root. | current working directory |
| TEST_GENIE_LLM_PROVIDER | No | LLM provider: anthropic, openai, or none. | auto-detect |
| TEST_GENIE_OPENAI_MODEL | No | Override OpenAI model. | gpt-4o-mini |
| TEST_GENIE_ANTHROPIC_MODEL | No | Override Anthropic model. | claude-haiku-4-5 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": true
} |
| resources | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| analyze_app_structureC | [mode: real] Static analysis of the project: screens, components, APIs, state. Auto-detects platform when not provided. |
| generate_scenariosC | [mode: real] Generate test scenarios from analyzed app structure. |
| create_test_planB | [mode: real] Build a test plan from stored scenarios with filtering / scheduling. |
| run_scenario_testB | [mode: hybrid] Run a stored scenario; real subprocess where possible, falls back to simulated. |
| run_simulationC | [mode: simulated] Random / sequential user-behavior simulation to find issues. |
| run_stress_testC | [mode: hybrid] Concurrency / load test against an endpoint or UI surface. |
| detect_memory_leaksC | [mode: real] Detect memory leaks, retain cycles, unclosed resources. |
| detect_logic_errorsC | [mode: real] Detect race conditions, null refs, state inconsistencies. |
| suggest_fixesC | [mode: real] Generate rule-based fix suggestions for detected issues. |
| confirm_fixB | [mode: real] Confirm / reject / modify a proposed fix prior to apply. |
| apply_fixA | [mode: real] Apply a confirmed fix. backup=true, validate=true by default. Supports dryRun. |
| rollback_fixA | [mode: real] Restore the pre-apply file content from the backup. |
| run_full_automationD | [mode: hybrid] Analyze → plan → execute → detect → suggest in one call. |
| run_iterative_fix_loopA | [mode: hybrid, headline tool] Self-healing test → fix → re-test loop with regression detection, auto-rollback, and resumeToken support. See docs/ITERATE_FIX_LOOP.md. |
| generate_reportC | [mode: real] Generate Markdown / HTML / JSON test automation report. |
| get_pending_fixesB | [mode: real] List fixes awaiting confirmation for the project. |
| get_test_historyC | [mode: real] Recent test executions for the project. |
| analyze_performanceC | [mode: real] Static performance analysis: rendering, computation, bundle. |
| analyze_code_deepC | [mode: real] Deep AST analysis: functions, complexity, hooks, issues. |
| generate_cicd_configC | [mode: real] Generate GitHub Actions / Jenkins / GitLab CI configuration. |
| diagnose_projectB | [mode: real, v3.1.0 headline] vibe-check the project — race conditions + security + memory + logic + performance in one parallel sweep. Returns prioritized findings + Markdown summary ready for chat. |
| detect_race_conditionsC | [mode: real, v3.1.0] Stand-alone race-condition detector — useState-after-await, missing AbortController, forEach-await, TOCTOU file ops, DispatchQueue races, Flow dispatcher mismatches. |
| detect_security_issuesB | [mode: real, v3.1.0] Stand-alone security scanner — hardcoded secrets (AWS/Stripe/GitHub/JWT), SQL/XSS/SSRF/eval injection, weak crypto, CORS misconfig, cookie flags, yaml.load. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| full-test-pipeline | Run the complete test-genie pipeline: analyze → plan → execute → iterate-fix-loop. Optimized for "fix until green or stuck". |
| diagnose-failure | Focused diagnosis of a single failing test, including a recommended next action. |
| vibe-check | One-command project diagnosis for vibe coders — race conditions + security + memory + logic + performance in 30s. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| iteration-logs-index | Index of all iterate-fix loop logs (most recent first). |
TDQS
Scored across 23 tools
Most tools have distinct purposes (e.g., analyze_app_structure vs analyze_code_deep vs analyze_performance). The main overlap is between the comprehensive diagnose_project and the individual detectors, but their descriptions clearly differentiate them. Slight potential confusion exists between analyze_code_deep and detect_logic_errors due to overlapping 'issues' coverage.
All tool names follow a consistent verb_noun pattern in snake_case (e.g., analyze_app_structure, run_scenario_test, generate_report). The naming convention is uniform across the entire set, with no mixed styles or vague verbs.
23 tools is on the heavier side and falls into the borderline range for a typical MCP server. While the domain is broad and each tool has a role, the redundancy between combined and stand-alone detectors (e.g., diagnose_project vs detect_race_conditions) adds some unnecessary weight.
The tool set covers a full lifecycle: analysis, scenario generation, planning, execution, detection, fixing, reporting, and CI/CD. However, there are minor gaps such as no explicit tools for listing or deleting stored scenarios/test plans, which may require workarounds.