v5.6.0 (2026-03-21)

Session memory enhanced, auto-gitignore for AI artifacts, and routing guide completeness. project-map.md is now injected directly into every session start by the hook — no instruction-following required. AI workspace files are automatically added to .gitignore the moment they're created. The routing guide now covers every user-invocable skill in the plugin.

New Features

project-map.md auto-injected at session start — The session-start hook now reads project-map.md from the working directory and injects its content directly into context, unconditionally, before the first turn. Previously, reading the project map relied on Claude following the entry sequence — if the task was classified as micro, or if the first message was conversational, the file might never be read. Now it's always in context. For maps ≤200 lines the full content is injected; for larger maps only the ## Critical Constraints and ## Hot Files sections are injected, with a note pointing to the full file.

Auto-gitignore for AI workspace artifacts — When Claude creates project-map.md, session-log.md, or state.md in a project, they are now automatically added to .gitignore under a # AI assistant artifacts section header. These are tooling artifacts — generated by the AI, not part of the project — and should never appear in git status as untracked files. The gitignore check runs in the PostToolUse hook for files Claude writes directly, and in the stop hook for session-log.md entries written by the hook itself. Idempotent: if the entry already exists, nothing is changed.

Changes

Routing guide now covers all user-invocable skills — premise-check and using-git-worktrees were missing from the routing guide in using-superpowers and had no coverage anywhere in the entry sequence. Both are now listed: premise-check at the top of the guide (run before brainstorming or planning when it's unclear whether work should exist at all), and using-git-worktrees before the implementation entries (run before implementation when the work needs branch isolation).

claude-md-creator added to routing guide with explicit bypass protection — CLAUDE.md creation was being classified as lightweight and implemented directly, bypassing the claude-md-creator skill that exists specifically for this task. The routing guide now includes an explicit entry for CLAUDE.md / AGENTS.md creation pointing to claude-md-creator, with a note that it applies at any complexity level. The lightweight action text now includes an exception: if a dedicated implementation skill exists for the task, invoke it — lightweight only skips workflow overhead, not implementation skills.

v5.5.0 (2026-03-20)

Reasoning gap prevention and fresh project memory setup. The AI now catches its own design and implementation gaps earlier, classifies tasks more accurately, and proactively offers to set up the memory stack before building anything in a new directory.

New Features

Fresh project gate — When you type "build", "create", "implement", or any creation-intent prompt in a directory with no project-map.md, the AI now pauses before starting and explains exactly what it will lose without the memory stack (re-exploring structure, re-reading known files, re-proposing rejected approaches, losing the "why" behind decisions). It offers to run git init and generate project-map.md in ~30 seconds before proceeding, or start immediately if you prefer. Previously this offer only appeared if git was absent — now it fires whenever no project-map.md exists, regardless of git status, so users who already have git initialized are no longer silently skipped.

Failure-mode check in brainstorming — Before any design can be approved, the AI must now state the top 2–3 ways the chosen approach could fail or not cover all cases. This is adversarial reasoning — actively trying to break the design — not a list of known assumptions. Critical failure modes (the design fails for a significant user scenario) must be fixed before proceeding; minor ones are documented as non-goals. This catches unknown assumptions at the design stage, where fixing them costs nothing, rather than discovering them after implementation.

Assumption externalization in writing-plans — The plan header now requires an Assumptions field listing what the plan rests on and what each assumption excludes ("Assumes X — will NOT work if Y"). Every task involving conditional logic now requires a Does NOT cover field stating which scenarios the condition excludes. If an excluded scenario should be covered, the task is revised before implementation begins. This catches known assumptions at the planning stage, complementing the adversarial failure-mode check in brainstorming.

Condition coverage check in verification — verification-before-completion now requires, as step 5 of its gate, that any change involving a condition or gate explicitly state what it does NOT cover before the task is marked done. If the answer reveals a gap that should be covered, it must be fixed before proceeding. This is the final catch in a three-stage adversarial pipeline: design → planning → completion.

Changes

Classification hard overrides — The complexity classification in using-superpowers now has a hard override section that is evaluated before the lightweight criteria. If any of four conditions are true (adds/modifies/removes a condition or trigger, affects user experience, modifies a shared dependency, introduces a new outcome), the task is immediately classified as full regardless of file count. This prevents lightweight anchoring on file count for tasks that have significant behavioral impact.

Lightweight articulation requirement — Before classifying any task as lightweight, the AI must now explicitly state in one sentence why each of the four lightweight criteria is satisfied. If any criterion cannot be clearly articulated, the task is classified as full. Combined with the hard overrides, this closes the gap where tasks with new conditional logic were being mis-classified as lightweight, skipping brainstorming and the design-stage failure-mode check.

v5.3.0 (2026-03-17)

Frontend design intelligence and documentation improvements. The frontend skill was completely rewritten from a 62-line checklist into a comprehensive design reasoning system, and several docs were updated to reflect the current state of the plugin.

New Features

frontend-design skill (complete rewrite) — The former frontend-craftsmanship skill has been renamed to frontend-design and rewritten from scratch. It now includes a 4-step design system generation framework that forces deliberate style, color, typography, and effects decisions before writing code. Adds a 25-style reference catalog (Minimalism through Cyberpunk), a 30-category industry design reference table mapping product types to recommended design directions, 8 common page structure patterns (dashboard, landing page, admin panel, etc.), 5-state UI state management (loading, error, empty, success, partial), frontend-backend integration patterns (API loading, optimistic updates, error boundaries, auth flows), dark mode implementation guidance, micro-copy and UX writing standards, and 10 priority quality standard categories covering accessibility, touch targets, performance, animation, forms, navigation, and charts. 353 lines, single file, zero dependencies.

Red team pipeline documentation — New docs/red-team-pipeline.md explains the end-to-end flow from code review through red team dispatch to auto-fix pipeline, including when each component fires, what the red team produces, how auto-fix processes findings, and merge blocking rules.

Frontend design documentation — New docs/frontend-design.md explains the skill's 7 capabilities with examples, what users can expect when prompting for frontend work, and how activation integrates with other Superpowers skills.

Changes

Testing documentation updated — docs/testing.md (now docs/testing-structure.md) was rewritten to reflect the actual 5-directory test structure: claude-code, skill-triggering, explicit-skill-requests, subagent-driven-dev, and opencode. Added the subagent hook scope test, fixed stale plugin name references, and added a quick reference section with copy-paste commands for every test suite.

AGENTS.minimal.md updated — Added missing premise-check skill reference.

frontend-craftsmanship → frontend-design rename — All references across 9 files updated: SKILL.md frontmatter, skill-rules.json, subagent-guard.js, README.md, AGENTS.minimal.md, using-superpowers routing, executing-plans, subagent-driven-development, and RELEASE-NOTES.md.