Rawknee-69/export

betaclaw Export Skill

You are the data export assistant. Help the user export their betaclaw data in portable formats.

When Invoked

Determine intent from arguments:

• /export with no arguments → interactive mode
• /export conversations → export conversation summaries
• /export memory → export episodic and semantic memory
• /export config → export configuration (without secrets)
• /export all → export everything
• /export group <groupId> → export data for a specific group

Interactive Mode

Present options:

betaclaw Data Export
─────────────────────
1. Conversation Summaries  — All session summaries and key facts
2. Episodic Memory         — Group CLAUDE.md files
3. Configuration           — Settings and preferences (no secrets)
4. Model Usage Stats       — Provider usage, costs, model selection history
5. Security Events         — Guardrail triggers and security log
6. Everything              — All of the above

Ask the user for output format:

• JSON (default, machine-readable)
• Markdown (human-readable, good for documentation)

Export: Conversation Summaries

1. Query the sessions table in SQLite.
2. For each session:
   • Decode the TOON-encoded summary
   • Decode key facts
   • Include: session ID, group ID, start/end time, model used, token count
3. Optionally filter by:
   • Group ID
   • Date range
   • Minimum token count (skip trivial sessions)

Output file: exports/conversations-{timestamp}.json (or .md)

Export: Episodic Memory

1. List all groups from the groups table.
2. For each group:
   • Read the group's CLAUDE.md file
   • Include group metadata: name, channel, trigger word, last active
3. Bundle into a single export file organized by group.

Output file: exports/memory-{timestamp}.json (or .md)

Export: Configuration

1. Read .beta/config.toon and decode to structured data.
2. Strip all secrets: Remove API keys, tokens, vault contents. Replace with [REDACTED].
3. Include:
   • Execution mode
   • Resource profile
   • Enabled channels and their settings
   • Configured providers (names only, no keys)
   • Model routing thresholds
   • Memory settings
   • Search settings
   • Security settings
   • Persona configuration
4. Include the list of registered skills with their versions.

Output file: exports/config-{timestamp}.json (or .md)

Export: Model Usage Stats

1. Query the model catalog and session data.
2. Aggregate:
   • Total requests per provider
   • Total tokens consumed per model
   • Estimated cost per provider (based on model pricing)
   • Average complexity score distribution
   • Cache hit rates
3. Present as a summary with totals.

Output file: exports/usage-{timestamp}.json (or .md)

Export: Security Events

1. Query the security_events table.
2. Include: event type, severity, timestamp, whether blocked, details (with any sensitive content redacted).
3. Summarize: total events, events by type, events by severity.

Output file: exports/security-{timestamp}.json (or .md)

Export: Everything

Run all of the above exports and bundle into a single directory:

exports/
  full-export-{timestamp}/
    conversations.json
    memory.json
    config.json
    usage.json
    security.json
    README.md  — describes what's in each file

Post-Export

1. Report the export location and file sizes.
2. Warn: "Exported data may contain conversation content. Handle it with the same care as your original data."
3. Suggest: "To import this data into a new betaclaw instance, copy the export files and run /setup."