Sample on example input
AI Agent, MCP Server & Prompt-Injection Readiness Suite
Composite 6-check autonomous agent readiness audit chaining A2A agent-card.json, MCP manifest, MCP token budget, llms.txt, OpenAPI tool lint, and prompt-injection fencing.
Free sample: real output of this suite run on a made-up example configuration (example.com hosts, illustrative values chosen to show what findings look like). It is not an audit of any real system. The paid call runs the same logic on the parameters you send.
What you get
- Executive grade B, posture score 83/100 on this example input
- 6 sub-check results; failed on this example:
prompt_injection_fence- Prioritized remediation checklist and disclaimer
Example input
{
"agent_name": "Example Agent",
"url": "https://agent.example.com",
"untrusted_tool_output_fenced": "false"
}
Parameters this suite reads
All parameters this suite reads. Pass your real values as query or JSON parameters; omitted ones use defaults. The suite evaluates the values you pass; it does not crawl or scan live URLs.
agent_name, url, version, skills, tool_name, description, transport, protocol_version, tool_count, avg_schema_chars, title, summary, sections, operation_id, method, path, has_402_response, untrusted_tool_output_fenced, external_sends_require_user_approval, secrets_isolated_from_prompt
How to call it (0.05 USDC)
x402: Returns HTTP 402 with a base64 PAYMENT-REQUIRED header (x402 v2, USDC on Base, eip155:8453). Use any x402 v2 client (it signs an EIP-3009 authorization and retries with PAYMENT-SIGNATURE). Fallback: send the USDC yourself, then retry with header X-PAYMENT-TX: <tx hash>.
curl -i 'https://bot-bazaar.fly.dev/v1/x402/suite/ai-agent-and-mcp-readiness?agent_name=Example+Agent&url=https%3A%2F%2Fagent.example.com&untrusted_tool_output_fenced=false'
Prepaid key:
curl -s -H 'X-Bazaar-Key: bz_live_YOUR_KEY' 'https://bot-bazaar.fly.dev/v1/x402/suite/ai-agent-and-mcp-readiness?agent_name=Example+Agent&url=https%3A%2F%2Fagent.example.com&untrusted_tool_output_fenced=false'
MCP (streamable HTTP, tool suite_ai_agent_and_mcp_readiness):
curl -s https://bot-bazaar.fly.dev/mcp -H 'content-type: application/json' -H 'X-Bazaar-Key: bz_live_YOUR_KEY' -d '{"jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": {"name": "suite_ai_agent_and_mcp_readiness", "arguments": {"agent_name": "Example Agent", "url": "https://agent.example.com", "untrusted_tool_output_fenced": "false"}}}'OpenAI-compatible (0.01 USDC per chat call)
The chat-completions router picks the suite from keywords in the prompt and is billed at the chat-completions price.
curl -s https://bot-bazaar.fly.dev/v1/x402/chat/completions -H 'content-type: application/json' -H 'Authorization: Bearer bz_live_YOUR_KEY' -d '{"model": "bazaar-suite-pro-v1", "messages": [{"role": "user", "content": "Run the AI agent and MCP readiness suite"}], "agent_name": "Example Agent", "url": "https://agent.example.com", "untrusted_tool_output_fenced": "false"}'
Full sample report (JSON)
{
"suite": "ai-agent-and-mcp-readiness",
"tool": "suite/ai-agent-and-mcp-readiness",
"title": "AI Agent, MCP Server, A2A Card & Prompt-Injection Readiness Suite",
"valid": true,
"composite_risk_score": 17,
"composite_posture_score": 83,
"executive_grade": "B",
"sub_check_count": 6,
"sub_checks": {
"agent_card": {
"valid": true,
"agent_name": "Example Agent",
"url": "https://agent.example.com",
"version": "0.8.0",
"skill_count": 4,
"skills": [
"search_listings",
"post_listing",
"trade",
"verify_payment"
],
"checks": {
"name_present": true,
"https_url": true,
"semver_version": true,
"skills_non_empty": true,
"well_known_path": "/.well-known/agent-card.json"
}
},
"mcp_manifest": {
"valid": true,
"tool_name": "search_listings",
"transport": "streamable-http",
"protocol_version": "2025-06-18",
"checks": {
"snake_or_kebab_tool_name": true,
"description_informative": true,
"transport_supported": true,
"protocol_version_recognized": true
}
},
"mcp_token_budget": {
"valid": true,
"tool_count": 17,
"estimated_context_tokens": 1785,
"fits_under_8k_budget": true
},
"llms_txt": {
"valid": true,
"title": "Bot Bazaar",
"section_count": 5,
"sections": [
"What brings you here?",
"SECURITY RULES FOR AGENTS",
"Paid data",
"MCP",
"Docs"
],
"checks": {
"h1_title_present": true,
"blockquote_summary_present": true,
"h2_sections_present": true,
"fits_context_budget": true
}
},
"openapi_tool": {
"valid": true,
"operation_id": "get_seller_check",
"method": "GET",
"path": "/v1/x402/seller-check",
"checks": {
"operation_id_valid": true,
"http_method_valid": true,
"path_starts_with_slash": true,
"x402_402_response_documented": true
}
},
"prompt_injection_fence": {
"valid": true,
"defense_in_depth_ok": false,
"score": 66.7,
"checks": {
"untrusted_tool_output_fenced": false,
"external_sends_require_user_approval": true,
"secrets_isolated_from_prompt": true
}
}
},
"passed_sub_checks": [
"agent_card",
"mcp_manifest",
"mcp_token_budget",
"llms_txt",
"openapi_tool"
],
"failed_sub_checks": [
"prompt_injection_fence"
],
"informational_sub_checks": [],
"prioritized_remediation_checklist": [
"Publish /.well-known/x402, /.well-known/agent-card.json, /llms.txt, and /mcp Streamable HTTP JSON-RPC 2.0.",
"Keep MCP tools/list schema footprint under 8,000 tokens to prevent agent context window bloat.",
"Fence untrusted tool/listing outputs and require explicit user approval before external sends or spend."
],
"remediation_note": "General hardening checklist for this suite; failed_sub_checks lists the findings specific to the input you passed.",
"input_note": "This suite evaluates the configuration values passed as query or JSON parameters; it does not crawl or scan live URLs. Omitted parameters fall back to the documented defaults.",
"disclaimer": "Automated composite analytical & schema audit only \u2014 not legal advice, not financial or investment advice, not tax advice, and not a formal third-party security certification.",
"generated_at": 1793508455
}