This repository was archived by the owner on Jun 9, 2026. It is now read-only.
feat(install): PR-01 — PAI-Install Electron GUI Installer (46 neue Dateien) - #68
Closed
Steffen025 wants to merge 215 commits into
Closed
feat(install): PR-01 — PAI-Install Electron GUI Installer (46 neue Dateien)#68Steffen025 wants to merge 215 commits into
Steffen025 wants to merge 215 commits into
Conversation
Complete plan for transforming PAI-OpenCode to PAI v4.0.3 structure: - Algorithm v1.8.0 → v3.7.0 - Flat skills → 11 hierarchical categories - Modular PAI/ directory structure - Full installer from v4.0.3 - Migration script for v2.x → v3.0 8 work packages defined with timelines, owners, and verification criteria.
Major updates to Epic: - Added DeepWiki research findings (6 questions, 100% success) - Documented OpenCode architecture: lazy loading, events, MCP, compaction - Updated Model Tier section with REAL production configuration - Added 15 agents with their Model Tier matrix (Engineer, Architect, etc.) - Clarified: Model Tiers are ALREADY in production, not future work - Strategic note: Custom binary = competitive advantage Key findings: - Model Tiers work TODAY in custom fork - DeepWiki confirmed: native lazy loading, 20+ plugin events, MCP integration - No need to build compaction (OpenCode has it native) Refs: DeepWiki API research, opencode.json production config
Added comprehensive Fabric analysis of Daniel Miessler's PAI videos: - extract_wisdom pattern applied to PAI v4.0 content - youtube_summary analysis of both videos - Key quotes and philosophy extraction - PAI v4.0 Feature Matrix with actions - Daniel's workflow habits (10 patterns) - One-sentence takeaway - Cross-reference with Release Notes Key findings: - Algorithm v3.7.0 is THE core (non-negotiable) - PAI is Infrastructure, not a Tool (Life OS positioning) - Human 3.0 vision = bridge to AGI - Euphoric Surprise = 9-10 ratings goal - 40 skills in 11 categories - 240+ Fabric patterns as knowledge base Refs: Fabric patterns (extract_wisdom, youtube_summary), PAI v4.0.2/v4.0.3 releases, Video analysis
Major WP restructuring based on research findings: KEY INSIGHTS APPLIED: - Model Tiers: Already production-ready (no dev, just use) - Lazy Loading: OpenCode-native (use skill tool, don't build) - Context Compaction: OpenCode-native (auto-handled) - MCP Skills: OpenCode-native (configure only) - Agent Swarms: Skip entirely (not available) WP CHANGES: WP1: Algorithm + Model Tier INTEGRATION (not building) WP2: Context MODERNIZATION using native lazy loading WP3: Event-Driven MIGRATION (hooks → native events) WP4: Skill reorganization (not rebuilding) WP5: MCP CONFIGURATION (not implementation) WP6: Voice FOUNDATION (arch only, not full V2V) WP7: Migration & Installer WP8: Testing & Release ADDED: - Critical Path analysis - Dependency graph - Effort estimates (hours) - Branch naming - Clear 'Key Insight' per WP - Verification criteria per WP Total estimated effort: 47-67 hours (was 40+) Now realistic based on native OpenCode features.
NEW: WP3.5 - Security Hardening (4-6h) - Prompt Injection detection module - Input sanitization layer - Output guardrails - Security event logging - Integration with PromptInjection skill for testing UPDATES: - WP3: Added Prompt Injection detection to plugin events - Dependencies: WP3 → WP3.5 → WP8 (Security now critical path) - Timeline: +4-6h, total 51-73h - Security features added to 'What We're Adding' Security Patterns: - detectPromptInjection() with regex patterns - Sanitize before LLM processing - Audit logging - Regular penetration testing Defense in depth: detect + sanitize + log + test
- Create SCOPE-BOUNDARY.md documenting project separation - Update EPIC-v3.0: Remove WP6 (Voice) - belongs to Open Arc - Restructure dependency graph without Voice/OMI work packages - Update timeline: 47-67h (reduced from 73h) - Add Scope section to README with clear IS/ISN'T table - Mark WP6 as MOVED TO OPEN ARC with explanation Voice-to-Voice and Ambient AI (OMI) are product features for jeremaiah-ai/openark, not the community PAI port. Refs: docs/SCOPE-BOUNDARY.md
- Update SKILL.md with Algorithm v3.7.0 content - Fix USER directory structure (Issue #29): Remove root .opencode/USER/ - Update docs/MIGRATION.md with correct canonical USER paths - Align with upstream PAI v4.0.3 structure
- Fix '4 red lines' header to '6 red lines' (4 original + 2 v1.3.0 additions) - Clarify PRD frontmatter schema: v1.0.0 canonical vs legacy, with field mappings - Remove non-existent PRDFORMAT.md references - Add language specifiers to code blocks (markdown, text, bash) - Fix CI secret scanning to exclude process.env references (false positive)
- Update PRD stub creation to use canonical v1.0.0 frontmatter schema - Remove isolated triple-backtick fence at line 439 - Update CI secret scan to exclude .env.example files - Fix process.env filter in secret scanning
- Add .opencode/node_modules/ to .gitignore - Update CI secret scan to exclude node_modules directories
- Add debug logging to show which files are being scanned - Show actual matches before filtering - Better error messages indicating which directory has issues
- Include YAML files in secret scan (*.yaml, *.yml) - Exclude BountyPrograms.json (contains public bug bounty URLs, not secrets) - Show only file:line in error output, not content (security best practice)
- Use grep -v to filter out BountyPrograms.json after the main search - This file contains public bug bounty URLs that are false positives
WP1: Port Algorithm v3.7.0 and Fix USER Directory Structure
- WP6: VoiceServer Foundation restored to PAI-OpenCode (TTS core only) - Voice-to-Voice/WebSocket remains in Open Arc - Agents location clarified: .opencode/agents/ (PAI 4.0.3 style) - 14 agent files: Algorithm, Architect, Engineer, Pentester, etc. - WP3: Added missing PAI 4.0.3 hooks list for porting - 12 hooks still need porting from upstream - 6 hooks already adapted in plugins/handlers/
- Add .opencode/PAI/ with correct upstream structure: - Algorithm/v3.7.0.md (383 lines) + LATEST symlink - SKILL.md (480 lines, Core only - not Algorithm!) - ACTIONS.md, AISTEERINGRULES.md, MEMORYSYSTEM.md - PAISYSTEMARCHITECTURE.md, PRDFORMAT.md, SKILLSYSTEM.md - THEDELEGATIONSYSTEM.md, THEHOOKSYSTEM.md, TOOLS.md - Tools/ (43 TypeScript tools from upstream) - USER/ (templates from upstream) - Update agents/ with full PAI 4.0.3 agent set (17 files) - Algorithm, Architect, Engineer, Pentester, QATester, etc. This corrects the previous WP1 which only updated .opencode/skills/PAI/SKILL.md Now we have the proper .opencode/PAI/ directory structure matching upstream. Closes: Revised WP1 implementation Related: Epic v3.0 WP1 requirements
- AddBg.ts: Replace exec with execFile (array args) to prevent shell injection - CI workflow: Add filters for example strings in documentation (False Positive prevention) More security fixes in follow-up commits
- AddBg.ts: exec→execFile (shell injection fix) - GetTranscript.ts: URL validation + execFileSync (shell injection fix) - RelationshipReflect.ts: exec→execFile + topic validation (shell injection fix) - BuildCLAUDE.ts: Remove .md.md double extension - RebuildPAI.ts: Remove .md.md double extension - CI workflow: Better false positive filtering
- FeatureRegistry.ts: Validate project name (path traversal fix) - SessionProgress.ts: Validate project + mkdirSync (path traversal + ENOENT fix)
- AddBg.ts: Require exactly 3 arguments (not at least) - FeatureRegistry.ts: Add JSON.parse error handling - FeatureRegistry.ts: Improve generateId with regex validation
- RebuildPAI.ts: Validate HOME directory, use .opencode/ paths - RebuildPAI.ts: Add file existence checks to loadAlgorithm - SessionProgress.ts: Add JSON.parse error handling - GetTranscript.ts: Add --save argument validation
- FeatureRegistry.ts: Use os.homedir(), validate status explicitly - FeatureRegistry.ts: Set completed_at to null (not delete) for schema compliance - GetTranscript.ts: 11-char YouTube IDs in examples, add /shorts/ support - RebuildPAI.ts: Update usage path to .opencode/, COMPONENTS_DIR existence check - RebuildPAI.ts: Robust frontmatter injection with regex, global replaceAll - SessionProgress.ts: os.homedir(), JSON.parse try/catch in loop - SessionProgress.ts: addBlocker only sets status if no resolution - SessionProgress.ts: decision requires all 3 args
- Skip args[3] if it starts with '-' (flag) instead of treating as status - Allows: feature-registry update <project> <id> --note ... - Without: fails with 'Invalid status: --note' SessionProgress handoff_notes: INTENTIONAL OVERWRITE (design decision) - Hand-off notes are internal algorithm system only - No user-created hand-offs, no residue expected
feat(pai): Revised WP1 - Correct PAI 4.0.3 Core Port
Replace 233KB static context with ~2KB minimal bootstrap using OpenCode-native lazy loading. Changes: - Add MINIMAL_BOOTSTRAP.md (~2KB) with core routing - Add CONTEXT_ROUTING.md with lazy loading documentation - Delete context-loader.ts (remove custom loader) - Update pai-unified.ts to use loadMinimalBootstrap() - Remove ContextResult from types.ts Verification: ✅ Bootstrap: 2KB (was 233KB) - 99% reduction ✅ Skills load on-demand via skill tool ✅ All 10 ISC criteria passing Related: EPIC-v3.0 WP2
Erweitert WP2 um System Rules + User Identity: Changes: - MINIMAL_BOOTSTRAP.md: Algorithm Essence + Steering Rules - Loader lädt jetzt: 1. Core Bootstrap 2. System AISTEERINGRULES.md (wenn vorhanden) 3. User Identity (ABOUTME, TELOS, DAIDENTITY) wenn existiert Prinzip: Nicht 'minimal möglich' sondern 'minimal nützlich' - AI muss den User kennen - Steering Rules müssen da sein - Alles andere lazy-loaded WP2 PR #34 ready for review
KRITISCH: Das System muss wissen, welche Skills existieren! Changes: - MINIMAL_BOOTSTRAP.md: Skill Discovery Registry hinzugefügt - 25+ Skills mit Triggern - 16+ Agent Types mit Invocation - Pattern-Matching Beispiel - CONTEXT_ROUTING.md: Dokumentation für Discovery-Mechanismus Problem gelöst: Vorher: System wusste nicht, dass z.B. 'Research' existiert Nachher: System hat Discovery Index → kann Skills identifizieren Bootstrap: 7KB (immer noch <25KB) Enthält jetzt: 1. Algorithm Core 2. System Steering Rules 3. User Identity 4. Skill Discovery Index ⭐️ WP2 PR #34 ready
Removed all German text from WP2 files: - MINIMAL_BOOTSTRAP.md: Rewritten in English - CONTEXT_ROUTING.md: Rewritten in English Content now fully international/English as required. All documentation, comments, and UI text is English only.
…consolidation Revert "docs(wp-n10): consolidate docs to final v3.0 state"
…ion-v2 docs(wp-n10): consolidate docs to final v3.0 state
# Conflicts: # .github/workflows/ci.yml # .opencode/PAI/ACTIONS/lib/pipeline-runner.ts # .opencode/PAI/ACTIONS/lib/runner.ts # .opencode/PAI/ACTIONS/lib/runner.v2.ts # .opencode/PAI/ACTIONS/pai.ts # .opencode/PAI/Tools/BannerMatrix.ts # .opencode/PAI/Tools/ExtractTranscript.ts # .opencode/PAI/Tools/FailureCapture.ts # .opencode/PAI/Tools/IntegrityMaintenance.ts # .opencode/PAI/Tools/OpinionTracker.ts # .opencode/PAI/Tools/PipelineMonitor.ts # .opencode/PAI/Tools/RemoveBg.ts # .opencode/PAI/Tools/SplitAndTranscribe.ts # .opencode/PAI/Tools/TranscriptParser.ts # .opencode/PAI/Tools/YouTubeApi.ts # .opencode/PAI/Tools/algorithm.ts # .opencode/PAI/Tools/pai.ts # .opencode/plugins/handlers/session-registry.ts # .opencode/plugins/handlers/skill-restore.ts # .opencode/plugins/handlers/voice-notification.ts # .opencode/plugins/lib/db-utils.ts # .opencode/skills/AudioEditor/Tools/Analyze.ts # .opencode/skills/AudioEditor/Tools/Polish.ts # .opencode/skills/Media/Art/Tools/GenerateMidjourneyImage.ts # .opencode/skills/Media/Art/Workflows/AdHocYouTubeThumbnail.md # .opencode/skills/Media/Art/Workflows/Comparisons.md # .opencode/skills/Media/Art/Workflows/Maps.md # .opencode/skills/Media/Art/Workflows/Visualize.md # .opencode/skills/Media/Remotion/ArtIntegration.md # .opencode/skills/Media/Remotion/Tools/Ref-timing.md # .opencode/skills/Media/Remotion/Tools/Ref-videos.md # biome.json # package.json
…R-12 Expands the v3.0 completion planning from 3 mechanical Claude categories to 8 comprehensive categories (paths, filenames, CLI calls, platform name, BuildCLAUDE.ts, claudeHome variable, session refs, JSONL transcript refs). New files: - CLAUDE-CLEANUP-PLAN.md: 452-line detailed per-file analysis with effort estimates, decision matrices, and PR assignments for all 8 categories Updated files: - V3.0-COMPLETION-PLAN.md: Added PR-12 section, expanded DoD checklist, updated statistics for 12 PRs - V3.0-RUNBOOK.md: Phase 1 expanded from 7 to 13 steps covering all 8 categories, added complete PR-12 execution section (branch, git commands, PR body, verification), expanded Definition of Done with 12 semantic cleanup checks, updated statistics for 12 PRs Key insight: The heaviest Claude files (THEHOOKSYSTEM.md 48 hits, TOOLS.md 25, MEMORYSYSTEM.md 20) are already identical on main and dev, requiring a dedicated PR-12 that branches from main directly.
…ing docs Discovery: THEPLUGINSYSTEM.md already exists at .opencode/skills/PAI/SYSTEM/ (363 lines, Jan 2026). Two THEHOOKSYSTEM.md files exist and both need deletion. Changes across all 3 planning docs: - THEHOOKSYSTEM: DELETE×2 (.opencode/PAI/ + .opencode/skills/PAI/SYSTEM/) - THEPLUGINSYSTEM: UPDATE (exists, needs 27-handler refresh) — not CREATE - Fix CodeRabbit findings: ```text tags on fenced blocks - Fix '6 .ts Dateien' → '5 .ts Dateien' (5 files in sed command) - Fix '11 PRs' → '12 PRs' where referring to total plan - Update effort estimates (1-2h update vs 3-4h create-from-scratch)
…x diagram branch, add sed portability, MD028, Mermaid diagram
…ansion docs(epic): expand v3.0 planning — 8-category Claude cleanup + PR-12
Contributor
|
Important Review skippedToo many files! This PR contains 264 files, which is 114 over the limit of 150. ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Run ID: ⛔ Files ignored due to path filters (36)
📒 Files selected for processing (264)
You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
📝 Coding Plan
Comment |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
PR-01 of 11 for PAI-OpenCode v3.0 finalization.
Adds the complete PAI-Install Electron GUI Installer — 46 new files enabling graphical installation and configuration of PAI-OpenCode without command-line knowledge.
Note: This branch is based on dev which contains all v3.0 work. The PAI-Install directory (46 files) is the focus of this PR. Other changes will be covered by PR-02 through PR-11.
CodeRabbit Review Focus
Review fuer Korrektheit, OpenCode-Konformitaet, und verbleibende Claude zu OpenCode Probleme.
Jira