acelm.ai / House (Public Specifications)
Docs > AceLM Documentation Index

AceLM Architecture & Lifecycle Documentation

Welcome to the technical documentation for AceLM (Agentic Coding Environment with Lifecycle Management).

AceLM brings professional engineering discipline to coding agents: defined requirements and architecture, evidence of work and quality, documented decisions, continuity for collaboration, and visibility into progress and resources. Experienced developers remain at the helm. AceLM is in development; capabilities still in design are identified explicitly.

Decision Precedence & Maturity

The documentation in this portal reflects the latest confirmed design decisions for AceLM. Detailed reconciliation notes, confirmed invariants, and open proposals are tracked in the Design Decisions & Alignment Record.

The documentation follows this precedence:

  1. Authoritative sources: Artifact Transition Catalog and Status Vocabulary & Terminal UX, including their explicit maturity boundaries.
  2. Downstream rules: lifecycle, metadata, isolation, assessment, and website references defer to those sources.
  3. Evidence register: alignment record reports corrections and checks; it does not create authority or finalize proposals.

Finalized Run/readiness semantics apply now; favored common lifecycle names await explicit adoption under the status migration contract.

Core Documentation Index

1. Overview & Architecture

2. Lifecycles & State Machines

3. Release Governance & Assessments

  • Gated Release Lifecycle: Release candidate definitions, candidate pinning, release gates, separate assessment verdict and action readiness, and post-release verification.
  • Sequential Release Assessment Cycle: Multi-pass evaluation algorithm (Pass 1 deep, Pass 2 light, Pass 3 convergence) and per-lens remediation inner loops.
  • 31-Lens Assessment Catalog: Complete catalog of 31 engineering lenses across Security, Reliability, Architecture, Performance, UX, and Compliance.

4. Research

Research about other products, the market, and how agent tools communicate with models lives in the

private acelm-research repository. Documents here cite

it by path. Material that moved there on 2026-10-11:

  • the OpenCode and Hermes Agent execution-flow analysis (docs/opencode-hermes-…-execution-flow.md) → agent-tools/findings/
  • docs/cursor-info/ → competitors/cursor/
  • docs/architecture-evaluations/acelm-client-interface-strategy.researched.md and its research prompt → market/

Visual Diagrams & Architectural Panels

Historical Baseline Lifecycle Panels

These preserved assets are historical shorthand, not the full current contract. Direct accepted-to-done arrows still require actual execution, evidence, and independent verification. Plan states do not define set states. The authoritative catalog and detailed diagrams govern all gates.

Detailed State Machine Diagrams

State Machine / Rule SVG Diagram Mermaid Source
Review Retries & Acceptance SVG Preview Mermaid
Verification Retries & Completion SVG Preview Mermaid
Completion Gates (E/V) SVG Preview Mermaid
Runtime Ownership & Failures SVG Preview Mermaid
Dependencies & Readiness SVG Preview Mermaid
Plan Item Evidence SVG Preview Mermaid
Workspace Isolation SVG Preview Mermaid
Release Lifecycle SVG Preview Mermaid
Release Assessment Gates SVG Preview Mermaid
Sequential Release Passes SVG Preview Mermaid

Core Architectural Invariants

  1. Persistent Git Records: All specifications, plans, decisions, and verification records travel with the Git repository.
  2. Execution Never Implies Verification: Fulfilling an E-* task does not satisfy a V-* requirement. Verification requires independent observation and reproducible test output.
  3. Adversarial Independence: Reviewing and verification tasks are performed in isolated agent contexts separate from authoring and execution sessions.
  4. Human Acceptance Authority: Agents propose; authorized humans accept. Release authorization is strictly bounded to concrete candidate revisions and target environments.