Ideas for Future Workflow Automation

June 8, 2026 · View on GitHub

Proposed Features

1. Auto-Open Browser on Startup

Current: User must manually navigate to http://localhost:3160/search after starting the engine.

Proposed: Automatically open browser and load search page when server starts.

// In engine/src/index.ts or startup script
server.listen(PORT, async () => {
  console.log(`Server running on port ${PORT}`);
  
  // Auto-open browser (macOS/Linux)
  if (process.platform !== 'win32') {
    const { exec } = require('child_process');
    exec(`open http://localhost:${PORT}/search`);
  } else {
    // Windows
    const { spawn } = require('child_process');
    spawn('start', ['http://localhost:3160/search'], { detached: true });
  }
});

Considerations:

  • User may not want browser to auto-start (privacy, distraction)
  • Should be configurable via user_settings.json
  • Respect user preference for manual launch

2. VS Code Integration / Auto-Detect Workspace

Current: User must manually add project path in the GUI.

Proposed: Detect active VS Code workspace and auto-add to watch list.

// On engine startup, check if launched from VS Code
const { env } = process;
if (env.VSCODE_CWD && env.VSCODE_PID) {
  // Add env.VSCODE_CWD to watched paths automatically
}

Alternative: VS Code extension that:

  • Watches for Anchor Engine running on localhost:3160
  • Automatically adds current workspace folder to watch list
  • Shows engine status in Status Bar
  • Provides "Start Watchdog" command

3. Incremental Index Correction

Current: Database rebuilds from scratch on startup (ephemeral design).

Proposed: Track file changes incrementally and update index as you code.

flowchart LR
    A[File Write] --> B[Watchdog Detect Change]
    B --> C[Extract Atoms]
    C --> D[Update Molecules]
    D --> E[Incremental Search Update]
    E --> F[Serve Queries]

Benefits:

  • No rebuild needed when editing a single file
  • Instant search updates as you type
  • Lower resource usage during development sessions

Challenges:

  • Maintaining consistency with atomic transactions
  • Handling deleted files and content removals
  • Managing provenance chain updates

4. Project-Specific Context Awareness

Current: All data goes into a single database.

Proposed: Automatically detect multiple projects in workspace and create separate contexts.

// Example structure:
~/.anchor/notebook/
├── project-a/      # Files from .qwenpaw/workspaces/P1/coding_projects/myapp
├── project-b/      # Files from .qwenpaw/workspaces/P2/coding_projects/api-server  
└── shared/         # Shared context (if any)

user_settings.json:
{
  "paths": {
    "notebook": "~/.anchor/notebook",
    "contexts": {
      "project-a": "~/coding_projects/myapp",
      "project-b": "~/coding_projects/api-server"
    }
  }
}

UI Implications:

  • Add "Switch Context" dropdown in search UI
  • Different color coding for different projects
  • Ability to run watchdog per-context

5. Incremental Corpus Backup

Current: Full corpus backup exports all molecules as YAML (takes time, requires I/O).

Proposed: Delta backups that only export changed records since last backup.

{
  "incremental_backup": {
    "last_full_backup": "2026-06-07T10:00:00Z",
    "changes_since": {
      "added": ["atom_abc123", "molecule_def456"],
      "modified": ["atom_ghi789"],
      "deleted": []
    },
    "checksums": { ... }
  }
}

Restore Process:

  • Full backup provides base snapshot
  • Incremental backups apply deltas in order
  • Much faster than full export for frequent backups

6. Live Preview Pane

Current: Search results require clicking to view full context.

Proposed: Side panel showing expanded molecule/atom content as you search.

block-beta
    column1: Search Results List
    column2: Context Expansion Area
    
    block:Result[Result Item]
        direction LR
        title["Record Title"]
        preview["Content Preview..."]
        expand[Expand ↗]
    end
    
    Result --> expand --> context[Full Atom Content]

Implementation:

  • Add "Expand" button to search results
  • Show full atom/molecule text in right panel
  • Support nested expansion (show related atoms)

7. Smart Tag Suggestions

Current: Tags must be manually assigned during ingestion.

Proposed: Auto-suggest tags based on content analysis.

// Example logic:
const analyzeTags = async (content: string) => {
  const techStack = ['typescript', 'nodejs', 'react', 'docker'];
  const keywords = techStack.filter(t => 
    content.toLowerCase().includes(t)
  );
  
  return [...new Set(keywords)]; // Deduplicate
};

Benefits:

  • Better searchability without manual effort
  • Consistent tagging across corpus
  • Can be overridden by user if needed

Priority Order

  1. Incremental Index Correction — highest impact for developer experience
  2. VS Code Integration — removes friction from setup
  3. Auto-Open Browser — convenience feature (low effort, high value)
  4. Project Context Awareness — enables multi-project workflows
  5. Live Preview Pane — improves search UX
  6. Incremental Backups — complements full backup system

Implementation Notes

Data Migration Path

When implementing incremental features:

  • Keep existing ephemeral database design
  • Add last_updated_at column to atoms/molecules tables
  • Track change timestamps for delta detection

Performance Considerations

  • Incremental updates should be <100ms per file
  • Avoid blocking UI during ingestion
  • Use background workers for heavy operations

Testing Strategy

  • Create "live-fire" test suite that:
    • Writes test files
    • Verifies immediate search availability
    • Checks provenance chain integrity