English Français العربية

IA-QA Eval & RAG — 38 tools

Official registry AI and agents

LLM & RAG evals: hallucination, eval contracts, semantic tests, tokens & model info. Free, no key.

Source code Documentation

Install

Add this to your client's MCP configuration (Claude Desktop: claude_desktop_config.json; Cursor: .cursor/mcp.json; VS Code and others use the same "mcpServers" shape).

Remote (streamable-http)
Claude Code 
claude mcp add --transport http ia-qa-eval-rag-38-tools 'https://www.ia-qa.com/mcp/eval'
JSON configuration (Claude Desktop, Cursor, VS Code) 
{
    "mcpServers": {
        "ia-qa-eval-rag-38-tools": {
            "type": "http",
            "url": "https://www.ia-qa.com/mcp/eval"
        }
    }
}

Documentation

From the project's README on GitHub, refreshed weekly. Links open on GitHub.

Contents
  1. IA-QA MCP server
  2. Endpoints
  3. Install
  4. What leaves your machine
  5. Packs

IA-QA MCP server

Free QA and LLM-testing tools for AI agents, served over MCP (Streamable HTTP). No account, no API key. Every tool returns a verdict an agent can act on: pass/fail, a score, a diff.

This repository describes the hosted server; it holds no server code. The live tool list is always GET https://www.ia-qa.com/mcp.

Endpoints

EndpointTools listedFor
https://www.ia-qa.com/mcpall 152clients with no tool limit
https://www.ia-qa.com/mcp/security30prompt injection, PII and secret scans, TLS/header/CORS audits, MCP server checks
https://www.ia-qa.com/mcp/eval38hallucination, eval contracts, semantic and RAG tests, tokens and models
https://www.ia-qa.com/mcp/data36JSON/YAML/CSV, schema validation, encoding, text and time utilities

Clients cap how many tools they load (VS Code Copilot 128 per request, Cursor about 40). A pack lists only its tools; for a tool outside it, add the full server or the pack that lists it — find_tool searches everything and says which pack lists each tool.

Install

Claude Code

claude mcp add --transport http ia-qa https://www.ia-qa.com/mcp
# or one pack:
claude mcp add --transport http ia-qa-security https://www.ia-qa.com/mcp/security

Cursor — .cursor/mcp.json (this project) or ~/.cursor/mcp.json (all projects) (use a pack: Cursor loads about 40 tools)

{
  "mcpServers": {
    "ia-qa-security": {
      "url": "https://www.ia-qa.com/mcp/security"
    }
  }
}

VS Code (GitHub Copilot, agent mode) — .vscode/mcp.json

{
  "servers": {
    "ia-qa": {
      "type": "http",
      "url": "https://www.ia-qa.com/mcp"
    }
  }
}

Claude Desktop — Settings → Connectors → Add custom connector, URL https://www.ia-qa.com/mcp (or a pack URL).

Any other MCP client — Streamable HTTP at one of the URLs above. More clients, and a prompt you can paste into any agent: https://www.ia-qa.com/mcp-server

What leaves your machine

Every call runs on ia-qa.com: whatever you pass to a tool is sent there — including a provider API key given to one of the few bring-your-own-key tools. The call journal stores no IP address: a monthly pseudonym (an HMAC, not reversible) to count distinct callers within a month, and each entry is deleted 400 days after the call. Rate limit: 60 requests per minute per IP.

For tools that must run next to your app (selector self-healing, app discovery), use the local packages instead — nothing is sent to ia-qa.com: @ia-qa/pal, @ia-qa/self-healing, @ia-qa/qa-discovery.

Packs

security — https://www.ia-qa.com/mcp/security

prompt_injection_scan · shield_analyze · guardrail_test · toxicity_scan · bias_detect · redact_pii · secret_scan · detect_secrets · security_headers_check · ssl_certificate_check · cors_test · cors_checker · cookie_security_audit · web_security_audit · validate_url · decode_jwt · generate_hmac · hash_text · generate_password · mcp_server_evaluate · mcp_server_health_check · mcp_schema_lint · validate_mcp_response · function_call_validate · validate_agent_trajectory · latency_benchmark · parse_http_headers · http_status_lookup · find_tool · rate_tool

eval — https://www.ia-qa.com/mcp/eval

get_testing_guidelines · run_eval_contract · run_semantic_tests · generate_eval_yaml · generate_ci_workflow · metamorphic_check · needle_haystack_generate · test_skill · hallucination_check · similarity_score · consistency_check · llm_output_validator · llm_format_check · llm_json_schema_check · response_quality_score · compare_responses · analyze_responses · prompt_test_suite · cot_analyzer · conversation_analyze · ab_test_report · embedding_similarity · rag_relevance_rank · rerank_evaluate · bm25_score · build_rag_prompt · count_tokens · token_budget_calculator · context_window_check · list_llm_models · model_info · compare_models · estimate_llm_cost · llm_generate · run_vlm_test_suite · extract_json_from_text · find_tool · rate_tool

data — https://www.ia-qa.com/mcp/data

format_json · json_schema_validate · json_diff · extract_json_path · flatten_json · json_to_yaml · yaml_to_json · xml_to_json · parse_csv · json_to_csv · mock_from_schema · json_schema_generate · openapi_validate · timestamp_convert · cron_parse · validate_email · regex_test · diff_text · text_stats · case_convert · generate_slug · base64_encode · base64_decode · url_encode · url_decode · generate_uuid · hash_text · number_base_convert · color_convert · lorem_ipsum · decode_jwt · env_parse · count_code_lines · lint_commit_message · find_tool · rate_tool

All 152 tools, by category: TOOLS.md.

Generated from the server's own definitions on 2026-10-26. The live list (GET https://www.ia-qa.com/mcp) wins if this page disagrees.

Frequently asked questions

What is IA-QA Eval & RAG — 38 tools?

LLM & RAG evals: hallucination, eval contracts, semantic tests, tokens & model info. Free, no key.

How do I install IA-QA Eval & RAG — 38 tools?

It runs as a remote server. In Claude Code: claude mcp add --transport http ia-qa-eval-rag-38-tools 'https://www.ia-qa.com/mcp/eval'. For other clients, paste the JSON configuration from this page into their MCP settings.

Does IA-QA Eval & RAG — 38 tools need an API key?

Its registry entry lists no required key or secret. Some features may still need an account with the service it connects to.

Is IA-QA Eval & RAG — 38 tools open source?

Yes. Its code is on GitHub under the MIT license.

Who maintains IA-QA Eval & RAG — 38 tools?

It is published by ia-qa-com. AI.tn lists it but does not maintain or endorse it.

From the official MCP Registry. Listing is not an endorsement: review a server's code and permissions before giving it access to your data.