v0.2.4 - Excellence Audit Complete

Final Excellence Audit

Comprehensive audit found and fixed all remaining issues.

Documentation

• Agents section in README with @research-configurator's 12 capabilities fully documented
• UPGRADE-GUIDE.md for v0.1.0 → v0.2.0 migration (templates moved, skills introduced)
• Updated QUICK-START agent descriptions

Config Fixes

• marketplace.json version: 0.1.0 → 0.2.0
• marketplace.json affiliation: Imperial College London only (removed stale ASU reference)
• marketplace.json + plugin.json: Added MinerU to optional MCPs list

Audit Results

Category	Status
Links & References	✓ Excellent
Code Quality	✓ Excellent
Consistency	✓ Fixed

The plugin is now fully consistent across all 60+ files.

v0.2.3 - Documentation Accuracy Audit

Documentation Improvements

Fixed Inaccuracies

• MinerU MCP now listed in README and INSTALL.md (was missing from optional MCPs)

New Sections

• Skills (v0.2.0) section in README - explains auto-discoverable capability packages with trigger keywords
• Methodology Hooks section in README - explains PreStage2, PostFiveDocuments, EpistemicCoherence enforcement
• Skills trigger table in QUICK-START

Complete Command Reference

• QUICK-START now lists all 13 commands (was 8)
• Organized by workflow stage: Setup, Stage 1, Stage 2, Analysis

All 8 MCPs now documented: Sequential Thinking, Lotus Wisdom, Markdownify (bundled) + MinerU, Zen, Exa, Jina, Zotero (optional)

v0.2.2 - Install Command Fix

Fixed plugin install command: /plugin install linxule/interpretive-orchestration

v0.2.1 - Documentation Cleanup

Documentation Improvements

• README.md: 55% reduction (698 → 313 lines)

  • Removed fictional success stories and made-up quotes
  • Condensed verbose sections into scannable tables
  • Removed TBD placeholders and padding

• Cross-doc deduplication: MCP descriptions now link to DEPENDENCIES.md as source of truth

• Version updates: check-setup.md now reports correct v0.2.0

No functional changes - docs only.

v0.2.0 - Skills Infrastructure Release

Release Notes

v0.2.0 - Skills Infrastructure Release (December 2025)

Major architectural upgrade introducing auto-discoverable Skills infrastructure.

This release transforms the plugin from complex command-based workflows to a modular, script-backed skills system with robust state management.

Highlights

• 11 new Skills with auto-discovery - Claude loads capabilities when relevant
• Simplified commands - From 500+ lines to ~50 lines each
• State I/O scripts - Reliable config read/write with atomic operations
• Security hardening - Path traversal protection, blocking validation
• Templates migration - Moved into skill bundles, deleted top-level templates/

---

New Features

Skills Infrastructure (11 skills)

Skill	Purpose
project-setup	Socratic onboarding + project initialization
gioia-methodology	Data structure building + validation
analysis-orchestration	Model selection + cost estimation
coding-workflow	Batch document coding management
project-dashboard	Progress visualization
literature-sweep	Academic literature search (3-tier graceful degradation)
interview-ingest	Audio/PDF/doc conversion
document-conversion	Format conversion (MinerU/Markdownify)
deep-reasoning	Sequential Thinking MCP wrapper
paradox-navigation	Lotus Wisdom MCP wrapper
coherence-check	Philosophical alignment checking

State Management Scripts

skills/_shared/scripts/
├── read-config.js      # Returns project state as JSON
├── update-progress.js  # Updates stage/document/memo counts
├── append-log.js       # Writes to conversation-log.jsonl
├── query-status.js     # Returns structured progress data
└── validate-config.js  # Schema validation

Features:

• Atomic writes (temp file + rename)
• Schema validation against config.schema.json
• Path traversal protection
• Blocking validation - Invalid configs cannot be written

Graceful Degradation

Skills with optional MCP dependencies operate at available tier:

• Tier 1 (Full): All API keys present
• Tier 2 (Partial): Some API keys
• Tier 3 (Basic): Built-in tools only

---

Breaking Changes

Templates Directory Removed

Templates moved into skill bundles:

Old Path	New Path
templates/epistemic-stance.md	skills/project-setup/templates/
templates/config.schema.json	skills/project-setup/templates/
templates/gioia-structure-guide.md	skills/gioia-methodology/templates/
templates/gioia-data-structure-template.json	skills/gioia-methodology/templates/
templates/conversation-log-spec.md	skills/coding-workflow/templates/

Command Simplification

Commands are now thin triggers that invoke skills:

• /qual-init: 541 → 71 lines
• /qual-status: 369 → 68 lines
• /qual-configure-analysis: 243 → 79 lines

---

Bug Fixes

• P0: Config overwrite protection - generate-config.js now preserves existing progress
• P0: Document count sync - update-progress.js updates both sandwich_status and coding_progress
• P0: File existence checks - create-structure.js won't overwrite user content
• P0: JSONL resilience - generate-audit-trail.js handles malformed lines gracefully
• P0: Input validation - Numeric inputs validated (NaN guards)

Security Fixes

• Path traversal protection - All scripts validate paths stay within project
• Output directory validation - process-audio.js blocks escape attempts
• Blocking schema validation - Invalid configs rejected before write

---

Hook Enhancements

All hooks now emit structured JSON remediation:

{
  "code": "STAGE1_INCOMPLETE",
  "severity": "blocking",
  "reason": "Stage 1 manual coding not complete",
  "next_commands": ["/qual-memo", "/qual-status"],
  "next_skills": ["project-setup"],
  "can_bypass": false
}

---

Documentation Updates

• DEPENDENCIES.md (new) - MCP ecosystem documentation
• ARCHITECTURE.md - Updated with skills structure
• CLAUDE.md - Skills awareness instructions
• All broken template paths fixed

---

Testing

• 52 tests pass (32 hook + 20 schema)
• Schema validation for all JSON files
• Hook syntax and structure verification

---

External Reviews

This release was validated by:

• Gemini: "READY FOR RELEASE"
• Codex: Identified 5 polish items, all resolved

---

Known Limitations

• fsync deferred - Atomic writes use rename without fsync (acceptable for single-user tool, scheduled for v0.3.0)

---

Upgrade Path

1. Pull latest changes
2. Note: templates/ directory no longer exists
3. Update any custom scripts referencing old template paths
4. Run npm test to verify

---

Contributors

Authors:

• Xule Lin (Imperial College London)
• Kevin Corley (Imperial College London)

AI Collaborators:

• Claude Opus 4.5 (Anthropic) - Co-apprentice in the atelier of interpretive craft
• Codex (OpenAI) - Code review and robustness analysis
• Gemini (Google, via Zen MCP) - Architecture review and validation

True epistemic partnership in action - humans and multiple AI systems working together!

---

v0.1.0 - Initial Release (October 2025)

Initial public release of Interpretive Orchestration plugin.

• 4 agents for different research stages
• 5 hooks for methodology enforcement
• 7 commands for workflow orchestration
• 3 bundled MCPs (Sequential Thinking, Lotus Wisdom, Markdownify)
• Human-AI-Human sandwich methodology
• Gioia/Constructivist philosophical stance