AIESv1.0 corpus

Open · Community driven · Vendor neutral

AI Engineering Standard

The open, vendor-neutral community standard for production AI delivery

Knowledge base and operating playbooks are shipped. What remains Draft is industry adoption: external pilots, shared governance, and inter-org critique — not inventing the practices.

Open the repository →Practitioner handbookView roadmap

Maintained in the open. Attribution: project steward · not a personal portfolio.

Why AI engineering needs a standard

Teams invent private dialects for prompts, agents, evals, and approvals. The result is unreproducible delivery, silent security gaps, and tools that cannot interoperate. AIES exists so organizations share one vocabulary, one set of gates, and one way to evidence production readiness — without locking to a single model vendor.

Trust comes from documentation, versioning, and community critique — not from declaring “the standard” on day one.

Standards by capability

Organized by engineering concern — not by a laundry list of frameworks.

AI SDLC / AIDLC
Stage gates, plans, HITL, and production prompts
Prompt engineering
Structured prompts, anti-patterns, templates
Context engineering
Minimal high-signal packages, progressive disclosure
Skills & agents
Procedures, tool-bounded actors, termination
Patterns
Planner, reviewer, release, hotfix, and more
Memory & knowledge
Hybrid RAG, lifecycle, poisoning defenses
MCP integrations
When to add tools — and how not to
Evaluation & LLMOps
Gates, judges, tracing, cost discipline
Governance
Enterprise controls, evidence, SoA semantics
Org playbook
Roles, operating model, assurance culture

Engineering principles

The constitution every AIES artifact should obey.

Full principles →
Human firstAI assistedReproducibleObservableTestableDeterministic where it mattersVendor neutralSecure by defaultContext rich, noise poorDocumentation driven

Adoption roadmap

Corpus is shipped. What remains is external adoption and shared governance.

Detailed roadmap →
  1. A1Corpus publish — prompts, skills, MCP, evals, AIDLC, patternsdone
  2. A2Packaging & discoverability for external contributorscurrent
  3. A3External pilots with published evidence notesnext
  4. A4Shared governance + non-steward RFC reviewersplanned
  5. A5Inter-assessor SoA reliability notesplanned
  6. S1Community-standard claim (adoption earned)future

Community & contributing

AIES improves through issues, PRs, pilot evidence, and disagreement on the record.

Contributing guideGitHubStandards portal