Reproducible fixtures

Examples use repository fixtures and the current reference scanner.

Every score or finding shown here is generated from a local fixture path and command recorded in the evidence file.

OpenAPI refund exampleMCP email example
Policymin-score 75 / fail-on criticalSourcesagentready-examples/commercial

OpenAPI bad fixture

Refund operation before remediation

Fixture: agentready-examples/commercial/openapi-refund-risk.bad.json

AgentReady OPENAPI scan
Score: 54/100
Status: Needs fixes
Operations/tools: 1
Risks: critical=1, high=4, medium=6, low=0

Top findings:
- [critical] AR002_MISSING_CONFIRMATION_BOUNDARY (missing_human_confirmation_flow) — refundOrder
- [high] AR001_UNBOUNDED_WRITE_ACTION (unbounded_parameter) — refundOrder
- [high] AR006_MISSING_DRY_RUN_OR_PREVIEW (irreversible_action) — refundOrder
- [high] AR005_SENSITIVE_DATA_EXPOSURE (sensitive_data_exposure) — refundOrder
- [high] AR007_OVERBROAD_TOOL_SCOPE (overbroad_permission) — refundOrder

Policy: FAIL
- score 54 is below --min-score 75
- 1 critical+ risk(s) detected

OpenAPI fixed fixture

Refund operation after contract tightening

Fixture: agentready-examples/commercial/openapi-refund-risk.fixed.json

AgentReady OPENAPI scan
Score: 84/100
Status: Minor fixes
Operations/tools: 1
Risks: critical=0, high=3, medium=0, low=0

Top findings:
- [high] AR006_MISSING_DRY_RUN_OR_PREVIEW (irreversible_action) — requestOrderRefundWithApproval
- [high] AR005_SENSITIVE_DATA_EXPOSURE (sensitive_data_exposure) — requestOrderRefundWithApproval
- [high] AR007_OVERBROAD_TOOL_SCOPE (overbroad_permission) — requestOrderRefundWithApproval

Policy: PASS

MCP bad fixture

Email tool before remediation

Fixture: agentready-examples/commercial/mcp-email-risk.bad.json

AgentReady MCP scan
Score: 63/100
Status: Needs fixes
Operations/tools: 1
Risks: critical=1, high=3, medium=4, low=0

Top findings:
- [critical] AR002_MISSING_CONFIRMATION_BOUNDARY (missing_human_confirmation_flow) — send_email
- [high] AR005_SENSITIVE_DATA_EXPOSURE (sensitive_data_exposure) — send_email
- [high] AR007_OVERBROAD_TOOL_SCOPE (overbroad_permission) — send_email
- [high] AR003_DESTRUCTIVE_OPERATION_AMBIGUOUS (mcp_dangerous_tool_weak_description) — send_email
- [medium] AR009_UNCLEAR_AGENT_INSTRUCTIONS (ambiguous_tool_description) — send_email

Policy: FAIL
- score 63 is below --min-score 75
- 1 critical+ risk(s) detected

MCP fixed fixture

Email tool after contract tightening

Fixture: agentready-examples/commercial/mcp-email-risk.fixed.json

AgentReady MCP scan
Score: 90/100
Status: AgentReady
Operations/tools: 1
Risks: critical=0, high=2, medium=0, low=0

Top findings:
- [high] AR005_SENSITIVE_DATA_EXPOSURE (sensitive_data_exposure) — send_confirmed_external_email
- [high] AR007_OVERBROAD_TOOL_SCOPE (overbroad_permission) — send_confirmed_external_email

Policy: PASS

Example limitations

These examples prove reproducibility of current fixtures only. They are not benchmarks, customer case studies or scientific validation.

TimeProofs AgentReady does not guarantee that an AI agent will never fail. It identifies structural risks that may cause AI agents to misuse APIs, tools or MCP servers.

AgentReady at a glance

AgentReady examples use repository fixtures and current scanner output; they are not customer evidence or benchmark validation. AgentReady is a product and candidate open standard for static pre-deployment analysis. It is not a runtime firewall, independent certification, official standards-body standard or guaranteed-safety system.

Availability
AgentReady Community is free and available. AgentReady Pro is planned and not purchasable.
Language
English is the current canonical language. No translated alternate route is published for this page.
Sources

AGENTREADY_MASTER_PLAN.mdGLOBAL_STANDARD_SITE_PROGRAM.mdSEO_GEO_AI_FIRST_REQUIREMENTS.mdSITE_COPY_GUIDE.mdCOMMERCIAL_FIXTURE_CI_GATE_BEHAVIOR.md