Skip to content

CHANGELOG.md

Latest commit

 

History

History
131 lines (98 loc) · 11.7 KB

CHANGELOG.md

File metadata and controls

131 lines (98 loc) · 11.7 KB

Changelog

v2.0.1.0 (2026-05-02)

Fixed

  • Plugin install actually exposes namespaced sub-skills. v2.0.0 promised /idstack:<skill> but ./setup symlinked the repo into ~/.claude/skills/idstack, where Claude Code only discovers the top-level dispatcher SKILL.md (one level deep). The 11 sub-skills under nested directories were never registered. Setup now installs to ~/.claude/plugins/idstack (the plugin discovery path), and the 11 skills live under skills/ per the plugin layout — so /idstack:needs-analysis, /idstack:pipeline, etc. show up in the slash command picker.
  • Circular self-symlink when cloning into the install target. The README's recommended one-liner clones into the install path, then runs ./setup. The script unconditionally ran ln -snf $REPO $TARGET even when $REPO == $TARGET, producing a broken self-symlink (~/.claude/skills/idstack/idstack -> ~/.claude/skills/idstack) inside the repo. Setup now detects the same-path case and skips the symlink step (the install is already in place).
  • Removed root dispatcher SKILL.md. The dispatcher only existed as a workaround for the broken symlink discovery — once the plugin install works, the namespaced skills replace it. The welcome-message logic moves into the slash command picker, where users see all 11 sub-skills directly.
  • Setup migrates legacy v2.0 installs. If ~/.claude/skills/idstack is a symlink, setup removes it. If it's a real directory (i.e., the v2.0 README had users clone there), setup leaves it in place but warns and tells the user how to remove it manually.

Changed

  • Repo layout: skill directories moved from repo root into skills/ subdirectory (required by the plugin format).
  • bin/idstack-gen-skills now reads templates from skills/*/SKILL.md.tmpl.
  • Default install path in preamble fallback, README, and CLAUDE.md updated to ~/.claude/plugins/idstack.

v2.0.0.0 (2026-04-20)

Added

  • Pipeline orchestrator. /idstack:pipeline chains all 8 design skills automatically, auto-skipping completed ones. Pause anytime, resume later. Your progress is saved in timeline.jsonl.
  • Cross-course intelligence. Learnings from one course now appear in another via ~/.idstack/global/learnings.jsonl. Keyword search with --keyword and --cross-project flags.
  • /idstack:learn skill. Search, list, delete, promote, and export project learnings. Promote local discoveries to the global store for cross-project reuse.
  • Course readiness dashboard. bin/idstack-status --readiness shows a pre-export gate: quality score, red-team critical findings, accessibility score, with pass/fail verdict. Integrated into /idstack:course-export.
  • Designer profile. Create ~/.idstack/profile.yaml with experience_level: novice|intermediate|expert. Skills adapt explanation depth to your expertise.
  • Manifest preferences. Schema v1.3 adds preferences section: verbosity, export_format, preferred_lms, auto_advance_pipeline.
  • Sub-agent architecture. On Claude Code, review skills dispatch parallel sub-agents for speed: /idstack:red-team (5 agents), /idstack:accessibility-review (2 agents), /idstack:course-quality-review (3 agents). Graceful sequential fallback on other platforms.
  • Spec review loop. /idstack:course-builder validates alignment via adversarial sub-agent after generating content. Reports "Review: N issues found, M fixed."
  • Claude Code plugin. .claude-plugin/plugin.json manifest for marketplace distribution. Install via claude plugin install idstack or traditional git clone && ./setup.
  • IDSTACK_HOME env var. All internal paths now use ${IDSTACK_HOME:-~/.claude/skills/idstack}. Set this to install anywhere.
  • New bin scripts. bin/idstack-learnings-delete, bin/idstack-learnings-promote for managing learnings programmatically.

Changed

  • Namespace refactor. All skills now invoked via /idstack:<skill> (e.g., /idstack:needs-analysis) instead of /<skill>. Avoids name collisions with other skill packages.
  • Setup creates single symlink. ./setup now creates only ~/.claude/skills/idstack (cleans up legacy individual symlinks automatically).
  • Manifest schema v1.3. Adds preferences section. Chained migration from v1.0/v1.1/v1.2 in one pass.
  • Preamble reads designer profile and preferences on every skill start.

v1.5.1.0 (2026-04-10)

Added

  • Bidirectional pipeline. All 9 skills now write back to the manifest. Previously 5 skills were read-only (course-quality-review, course-builder, red-team, accessibility-review, course-export). Now every skill contributes to the shared project state, so downstream skills get richer input.
  • Score trending. Run /course-quality-review multiple times and see your score improve: "Score: 78/100 (+16 since last review)." Current score in the manifest, history in the timeline. One source of truth per data point.
  • Export readiness info. /course-export now shows quality, red-team, and accessibility scores before export. Informational only, never blocks.
  • WCAG 2.1 AA depth. /accessibility-review expanded from 8 to 20+ evidence citations with full WCAG success criteria. Course-specific guidance for videos, quizzes, forums, PDFs, and simulations.
  • Red-team evidence grounding. /red-team expanded from 12 to ~45 evidence citations. All 5 adversarial dimensions now cite their research.
  • Chained schema migrations. bin/idstack-migrate refactored to support migration chains (1.0→1.1→1.2). Users on any version get upgraded in one pass.
  • Migration test fixtures. New test/fixtures/ with v1.0 and v1.1 manifest fixtures. 11 new migration assertions in smoke-test.sh.

Changed

  • Manifest schema bumped to v1.2 (additive only, no breaking changes).

v1.5.0.2 (2026-04-06)

Fixed

  • Security: Removed silent auto-update that pulled and executed code from GitHub without user review. Skills now notify when updates are available. Users update manually with git pull && ./setup.

v1.5.0.1 (2026-04-04)

  • Landing page now shows development updates and roadmap at idstack.org. Timeline with recent releases (dates as primary identifiers, not version numbers) plus "Coming soon" items and a "Where we're headed" narrative. Non-technical visitors can see what's new without navigating GitHub.
  • Updated architecture description on landing page to reflect course memory (timeline, learnings).
  • Removed stale architecture ASCII diagram from landing page (replaced with prose description).

v1.5.0 (2026-04-04)

  • Course memory. idstack now remembers your design sessions. Each skill logs what it did to .idstack/timeline.jsonl, and the next session starts with a welcome-back message: quality score trend, last skill run, suggested next step. Your course context compounds across sessions instead of starting from scratch.
  • Learnings. When a skill discovers something specific to your project (Canvas uses custom rubric formatting, SCORM packages from Rise need special handling), it stores the discovery in .idstack/learnings.jsonl. Future sessions surface relevant learnings automatically.
  • Pipeline guidance. Skills now suggest the next step based on what you've already completed, replacing inconsistent static "Next step:" text with data-driven recommendations.
  • bin/idstack-status prints a course health dashboard: skills completed, quality score trend, dimension breakdown, and suggested next skill.
  • Template system. Shared preamble code (update check, manifest check, context recovery) is now maintained in one place (templates/preamble.md). Skills use .tmpl files with a {{PREAMBLE}} placeholder. Run bin/idstack-gen-skills to regenerate.
  • python3 recommended. Course memory features work best with python3 (score trends, search filtering, JSON validation). Without it, basic timeline logging still works via bash fallback.

v1.4.1.1 (2026-04-03)

  • Setup now tells you to restart Claude Code if it's already running, so skills appear immediately
  • Post-install message routes you to the right starting skill: /course-import if you have an existing course, /needs-analysis if starting fresh

v1.4.1 (2026-04-02)

  • Landing page live at idstack.org with getting started guide, terminal demo, skill pipeline diagram, and evidence grid across 11 research domains
  • Skills check for updates and notify when a new version is available.
  • SEO: Open Graph and Twitter card tags, JSON-LD structured data, OG image, and favicon for link previews when shared on social media

v1.4.0 (2026-04-01)

  • You can now import SCORM 1.2 and 2004 packages directly into /course-import. Works with Articulate Rise, Storyline, Adobe Captivate, Lectora, iSpring, and any SCORM-compliant authoring tool. Extracts course structure, objectives, and assessments from the imsmanifest.xml.
  • You can now export to SCORM 1.2 packages from /course-export. Produces a standard SCORM ZIP that works with every LMS and corporate training platform.
  • PDF and document file import added to /course-import for Articulate Rise exports and course syllabi

v1.3.0 (2026-04-01)

  • You can now run /accessibility-review to audit your course for WCAG 2.1 AA compliance and Universal Design for Learning (UDL 3.0). Two-tier output: "Must Fix" for legal accessibility violations, "Should Improve" for inclusive design recommendations. Scores accessibility 0-100.
  • You can now run /red-team to stress-test your course design. Five adversarial dimensions: alignment gaps, evidence verification, cognitive load analysis, learner persona simulation, and prerequisite chain integrity. Produces a confidence score so you know how solid your design is before you export.
  • New evidence domain added: Domain 11 (Accessibility & Universal Design for Learning) with 9 citations covering WCAG, UDL Guidelines 3.0, and differentiated instruction research
  • Pipeline extended: /accessibility-review and /red-team sit between /course-quality-review and /course-export as the final quality gates
  • Smoke tests expanded from 46 to 58 checks covering all 9 skills

v1.2.1 (2026-04-01)

  • You can now submit feedback and feature requests without a GitHub account via a Google Form
  • Feedback link added to every skill, so you can share thoughts right after using one
  • README intro softened for a warmer, more inviting tone
  • Multi-platform install (Gemini CLI + Codex CLI) added to roadmap as a P2 item

v1.2.0 (2026-03-31)

  • Cross-domain quality checks in /course-quality-review: cognitive load, multimedia principles, feedback quality, and expertise reversal flags
  • Manifest versioning with automatic schema migration (bin/idstack-migrate)
  • Fix-link integration so quality review recommendations point you to the skill that fixes each issue
  • Quick-win prioritization ranks recommendations by effort vs impact
  • Score trending tracks your course quality across review sessions
  • Shareable quality report and per-category scoring breakdown
  • Pipeline reordered so /course-quality-review runs before /course-export
  • Auto-create GitHub Release on tag push

v1.1.1 (2026-03-31)

  • VERSION file added for tracking
  • Branch protection and CONTRIBUTING.md for contributors

v1.1.0 (2026-03-30)

  • 4 new skills: /assessment-design, /course-builder, /course-export, /course-import
  • Full pipeline from import to export now works end to end
  • Automatic update check in all skills (tells you when a new version is available)
  • README rewritten with install-by-pasting flow and live demo walkthrough

v1.0.0 (2026-03-30)

  • Initial release with 3 skills: /needs-analysis, /learning-objectives, /course-quality-review
  • Evidence base from peer-reviewed research across 11 domains
  • Shared project manifest (.idstack/project.json)
  • Evidence tier citations (T1-T5) on every recommendation