Skip to content

Changelog

Audience: Humans (project changelog)

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Versioning policy for this toolkit: - Major — Breaking changes to the mental model, skill interfaces, or reading order - Minor — New skills, new features, new output adapters - Patch — Bug fixes, documentation improvements, token optimizations

Unreleased

[2.6.0] - 2026-03-04

New Skills (3)

roadmap-building — Generates full phased implementation roadmaps from architecture analysis, TOGAF Phase E/F output, or standalone requirements. Produces initiatives, ADRs, risk register, epic/ticket breakdown with complexity sizing (XS–XL) and team assignments, and a machine-readable roadmap tracker. Integrates with TOGAF Phases E, F, and G.

roadmap-analysis — Seven-mode deep-dive analysis for an existing roadmap: (1) Expand an initiative into a week-by-week implementation plan with rollback triggers; (2) Identify research questions that need structured investigation; (3) Score complexity across technical, organisational, and risk axes with critical path; (4) Identify spike candidates with Go/No-Go criteria; (5) Generate itemised effort estimates with risk buffers; (6) Export roadmap packages to PDF; (7) Plan team composition and financial model by phase.

coding-profile — Reads real repository code and generates per-stack coding skill files agents can load before writing code. Captures naming conventions, async style, testing framework and patterns, error handling strategy, API design, observability setup, and review standards — all from actual code, not generic best practices. Groups multiple repos by stack and merges them into one profile, flagging divergences explicitly.

Quick Start & Onboarding

Updated the one-prompt Quick Start to use .quantum-toolbox as the canonical submodule directory name (replacing the legacy .ai-toolkit). The setup script, AGENTS.md template, and CONTEXT.md template all align on this name now.

Added a skill-first rule to both templates: before starting any task, agents check skills/_index.md for a matching skill and follow it if one exists. This is surfaced as a callout in AGENTS.md and as a prominent rule in CONTEXT.md, making it harder to miss.

Quick Start step 5 added: Read skill catalog — ensures the skill index is always loaded at session start.

Core Tooling & Conventions

Added > **Audience:** tags to all files that were missing them (AI_TOOLKIT_CONTEXT.md, CONTRIBUTING.md, SECURITY.md, CHANGELOG.md, CODE_OF_CONDUCT.md, specs/README.md, specs/ROADMAP-TRACKER.md). Every file in the toolkit now declares its intended reader.

docs/skills/README.md simplified: removed duplicated skill tables that were drifting out of sync with skills/_index.md. The file now points to the index as the single source of truth.

core/workflows.md updated: presentation, release-notes, sprint-plan, and review-output moved to a new Output Generation category in Skill Discovery — they were previously buried under Development Workflows or missing entirely.

.gitignore updated to exclude .claude/settings.local.json (user-specific path, not shareable).

Open-Source Readiness

  • CODE_OF_CONDUCT.md — Contributor Covenant
  • CONTRIBUTING.md — guide for building skills, adjusting the mental model, PR workflow
  • SECURITY.md — vulnerability reporting policy
  • LICENSE — MIT license
  • Mental model override mechanism (architecture-thinking.local.md) with template
  • README.md: "How It Works" section explaining the mental model as engine
  • AGENTS.md: updated reading order to include mental model override step

Improvements & Optimisations

  • Session-start token cost reduced from ~5,006 to ~4,620 tokens (~8%) by removing duplicated skill tables from docs/skills/README.md that were maintained separately from skills/_index.md
  • Skill Discovery in core/workflows.md now covers all invokable skills — 4 skills (release-notes, sprint-plan, review-output, presentation) were previously missing from the section agents use to list capabilities
  • specs/roadmap-skills-spec.md expanded with roadmap tracker format, epic/ticket breakdown, XS/S/M/L/XL complexity scale, and team assignment rules — used as the build spec for the new roadmap skills
  • skills/manifest.yaml and skills/_index.md kept in sync with full metadata for all new and existing skills
  • ROADMAP-TRACKER.md reflects all completed items across optimisation and new skills tracks

2.5.0 - 2026-02-08

Added

  • Token baseline measurement (specs/TOKEN-BASELINE.md)
  • Roadmap skills tracking in skills/_index.md and manifest.yaml
  • Expanded glossary with roadmap and financial terms

Changed

  • Consolidated roadmap tracker with optimization track
  • 48% reduction in session-start token cost (9,618 to 5,006 tokens)
  • Eliminated redundancy across core files

2.4.0 - 2026-02-06

Added

  • PDF report skill — structured document generation
  • Presentation skill moved to output adapters category
  • Skills documentation hub (docs/skills/README.md)
  • Skill discovery workflow for improved onboarding
  • Roadmap-analysis modes: team planning, financial estimation
  • Simplified startup prompt

Changed

  • Replaced "resources" terminology with human-centric language throughout
  • Directory restructure for clean submodule usage

2.3.0 - 2026-02-05

Added

  • Architecture Synthesis skill — middle-out architecture from diagrams and specs
  • Core Architecture output adapter — standardized analysis output format
  • Change management communication plan
  • Fitness Functions skill — quantitative architecture governance
  • Migration planning integration
  • Presentation skill — end-to-end slide deck delivery
  • NIS 2 EU Directive compliance framework for security-analysis
  • Excalidraw as 4th diagram output format
  • TOGAF Preliminary Phase skill

2.2.0 - 2026-02-04

Added

  • TOGAF ADM Phases E, F, G, H — complete ADM coverage
  • Roadmap for core architecture output and enhancements
  • Excalidraw and presentation generation roadmap
  • Tooling research for diagrams and presentations

2.1.0 - 2026-02-03

Added

  • Nonfunctional-analysis skill
  • Structurizr authoring skill — C4 model generation
  • TOGAF Core ADM skills (Phases B, C, D)
  • Security-analysis skill
  • Analysis refactoring (Phases 1-4) — unified analysis pipeline

Changed

  • Comprehensive README update with quickstart prompts
  • Renamed to "AI Agent Toolbox for Architects"
  • Added AI_TOOLKIT_CONTEXT.md with full repository context

2.0.0 - 2026-02-01

Added

  • Architecture thinking foundation (core/architecture-thinking.md) — the mental model engine
  • TOGAF skill family index
  • Arch-analysis structured output directory
  • Diagram format options and AI file exclusions
  • TOGAF vision checklist

Changed

  • Arch-analysis expanded with Dependency Health Check, Data Flow Mapping, Error Handling workflows

1.0.0 - 2026-01-29

Added

  • Git submodule architecture — toolkit designed for embedding
  • Tiered skills system (core + optional)
  • Arch-analysis skill — codebase architecture analysis
  • Todo-workflow skill — structured task management
  • Root README and architecture analysis guide
  • Skill index (skills/_index.md)

0.1.0 - 2026-01-18

Added

  • Initial AI agent setup with instruction files
  • Pre-work checklist and autonomous development loop
  • Core workflows and coding guidelines