Ai Sdlc

This open source framework orchestrates AI coding agents from issue to merged pull request. It treats software delivery as a series of well framed decisions: an idea is captured as a work item with explicit open questions, the operator resolves those questions until a Definition of Ready gate is satisfied, and an autonomous orchestrator then walks the dependency graph, dispatches developer subagents in isolated worktrees, runs reviewers in parallel and merges through declarative quality gates. Pipelines, policies and autonomy levels are declared as YAML resources with a spec and status split. It is aimed at engineering teams already using agents such as Claude Code, Codex, Cursor, Copilot or Aider who want governance around them. Cross harness review enforces that the agent that implemented a change cannot also review it, each change is sealed with a signed DSSE attestation, and quality gates escalate from advisory to soft to hard mandatory according to the compliance posture. An operator terminal interface shows decisions, the dependency graph and analytics, and a Claude Code plugin supplies hooks, commands, review agents and an MCP server. The vendor also targets regulated sectors such as financial services, with audit trails and risk tier classification mapped to the EU AI Act. The Community edition is free under the Apache 2.0 license, with no limits and community support through GitHub Issues. Enterprise pricing is not published and is handled through sales; it adds a managed settings generator for MDM, GPO and plist deployment, centralized audit hooks, SIEM export to Splunk, Datadog, Elastic and Sentinel, SSO and SAML, compliance reports for SOC 2, ISO 42001 and the EU AI Act, on premises deployment and dedicated support with an SLA. The specification is still versioned v1alpha1, so its resource formats should be treated as early and subject to change.