RFC-XXXX: [Feature / Standard Title]
- Start Date: YYYY-MM-DD
- Author(s): [Name / GitHub Handle]
- Status: Draft | Under Review | Approved | Rejected
- Target Area: Prompt Engineering | MCP Contracts | Evaluation Gates | AIDLC Stage Gates | LLMOps
Summary
Brief 2–3 sentence explanation of the proposed change or specification.
Motivation
Why are we doing this? What problem does it solve in production AI delivery? What are the failure modes or security gaps if this standard is omitted?
Detailed Specification
Provide the exact technical specification:
- Schema Definitions: JSON Schema, TypeScript types, or YAML specifications.
- Harness Rules: Specific rules for
AIES.md,AGENTS.md, or.cursor/rules. - Runtime Expectations: Expected behavior, input/output validation, and error handling.
Evidence & Verification
How will this standard be verified in CI/CD?
- Automated test scripts or
aies auditlinter checks. - Artifacts required to evidence compliance.
Alignment with Evolving Standards
Which external specifications does this align with?
- NIST AI RMF 1.0 (e.g., MANAGE 1.2, MEASURE 2.1)
- ISO/IEC 42001 (AIMS Controls)
- OWASP Top 10 for LLM Applications (e.g., LLM01 Prompt Injection, LLM07 Insecure Plugin Design)
- Model Context Protocol (MCP)
Drawbacks & Backward Compatibility
- Are there breaking changes?
- What is the migration path for existing AIES adopters?
Version: AIES v1.0.0✏️ Edit this page on GitHub