Changelog

Version history

A concise record of meaningful changes to the manifesto, lifecycle, operating model, and supporting delivery guidance.

Back to manifesto Read V1.5

September 2026

V1.5

Practical governance and competent oversight

  • Expanded to five principles, with minimum necessary autonomy and evidence-based authority.
  • Made the requirements quality gate incremental while keeping it mandatory before implementation.
  • Added autonomy contracts, runtime enforcement, memory governance, and the Behavioral Evaluation Skill.
  • Clarified recovery, operational value, and criteria for reducing autonomy or retiring systems.
  • Required trained human reviewers with evidence, sufficient time, and effective authority to intervene.
  • Added an enterprise adoption guide with a worked refund-assistance example.
  • Expanded tooling examples with Temporal, OpenBao, and OpenMetadata; listed Phoenix as an alternative and LangGraph as optional, and clarified policy enforcement and tool scope.

Technical Standard layout

Added a vendor-neutral tool-family map and a dedicated LLM gateway comparison, with open-source options first, multi-provider routing, and input/output token policies.

Introduced readable typography, open sections, a fixed desktop index, a collapsible mobile menu, and responsive tables across all four languages and the changelog.

September 2026

V1.4

Enterprise adoption

  • Strengthened ADLC's positioning as behavioral lifecycle governance that extends SDLC beyond code-centric delivery.
  • Added a minimum implementation baseline and an ADLC conformance checklist.
  • Added an SDLC-versus-ADLC comparison and clarified accountable human responsibilities.

Knowledge and evidence

  • Renamed RAG Governance Agent and Skill to Knowledge Governance Agent and Skill.
  • Expanded governance to RAG, prompts, shared knowledge, context endpoints, access rules, traceability, and behavioral regression evidence.
  • Improved the evidence and tooling guidance for enterprise implementation.

Resources and interface

  • Organized external references into governance, protocols, observability, practices, and optimization.
  • Added NIST, ISO/IEC 42001, MITRE ATLAS, OWASP, MCP, A2A, and OpenTelemetry GenAI resources.
  • Improved desktop and mobile table layouts for long enterprise content.

June 2026

V1.3

Documentation channels

  • Clarified that human documentation and agent context are connected but separate artifacts.
  • Positioned Confluence, SharePoint, Notion, GitBook, Backstage TechDocs, Read the Docs, and similar systems as human-facing documentation channels.
  • Added governed Agent Context Endpoints for AI agents, including MCP servers, llms.txt files, retrieval indexes, and versioned context packs.
  • Added examples of agent-context tooling, including Context7, GitMCP, MCPDoc, mcp-documentation-server, and custom MCP servers built with open MCP SDKs.

Responsive layout

  • Improved the fixed header for compact, readable navigation.
  • Made the operating model table readable on mobile by transforming rows into labeled cards.
  • Updated the site stylesheet cache key for the V1.3 release.

May 2026

V1.2

Knowledge governance

  • Added Knowledge Governance / RAG Governance across the manifesto, website, lifecycle, evidence model, shared agents, and shared skills.
  • Introduced the principle: in agentic systems, knowledge is behavior.
  • Clarified that RAG sources, prompts, shared skills, policies, examples, and tool instructions can change agent behavior even when application code does not change.
  • Added the RAG Governance Agent and RAG Governance Skill.

Evidence and project quality

  • Extended Test, Deploy, and Operate to cover prompt, RAG, model, tool, and orchestration changes.
  • Added traceability from requirement to knowledge source, agent behavior, test evidence, and release.
  • Added CI, local link validation, Makefile checks, repository attributes, README badges, and localized README files.

May 2026

V1.1

Lifecycle and practice

  • Added the ADLC in Practice section to make the manifesto more concrete.
  • Introduced a React Flow lifecycle diagram with animated double-loop flow.
  • Made lifecycle nodes clickable so each step links to its explanation.
  • Clarified the Requirements Quality Gate as the mandatory entry point into ADLC.
  • Repositioned Orchestrate as a central cross-cutting coordination layer.
  • Highlighted human-in-the-loop checkpoints in the diagram and lifecycle guidance.

Operating model and governance

  • Reworked the ADLC Operating Model into a clearer governance matrix.
  • Clarified that ADLC extends SDLC rather than replacing it.
  • Reframed external positioning as manifesto language instead of commentary.
  • Added shared agents for documentation, PR governance, release notes, traceability, and operational readiness.
  • Added shared skills for documentation, release notes, architecture, infrastructure, and CISO-led security.

SEO and sharing

  • Strengthened canonical, hreflang, robots, sitemap, Open Graph, and Twitter metadata.
  • Added a stable Open Graph image for LinkedIn and social previews.
  • Set the public domain to adlcmanifesto.org across SEO metadata.
  • Added this changelog page and exposed it through the version badge and footer.

Content and interface

  • Improved multilingual content across English, Italian, Spanish, and French.
  • Refined responsive layout, menu readability, and PaperCSS visual consistency.
  • Protected the contact email from simple spam scraping while keeping it usable.
  • Added a standalone shared skills markdown document for reuse outside the site.

April 2026

V1.0

Initial public version of the ADLC Manifesto: core principles, lifecycle framing, multilingual pages, PaperCSS visual direction, and the first definition of the ADLC as a governed, tool-agnostic lifecycle for agentic delivery.