garrettroi/skills/hive_mind

Hive Mind — Shared Memory System

The Hive Mind is the team's shared knowledge base stored in Redis. All agents can read from it. All agents can submit lessons to it. Only the Lexi curates and distributes knowledge.

How to Use

Reading Knowledge (All Agents)

Search the hive mind for lessons relevant to your current task:

python3 /app/skills/hive_mind/hive_search.py --query "your search terms" --agent "your_name"

This searches across all categories using keyword matching and returns the most relevant lessons. Always do this before starting a new task (see Boot Instructions in your SOUL.md).

Check your personal knowledge feed for lessons the Lexi has routed to you:

python3 /app/skills/hive_mind/hive_feed.py --agent "your_name"

Submitting Lessons (All Agents)

When you discover something useful during a task, submit it to the Lexi for review:

python3 /app/skills/hive_mind/hive_submit.py --agent "your_name" --title "Short descriptive title" --content "The full lesson content" --tags "tag1,tag2,tag3"

Tags should include relevant categories like: sales, real_estate, dj, photo_booth, cana, marketing, social_media, ads, automation, finance, client_relations, content, research.

Managing Knowledge (Lexi Only)

Review the intake queue:

python3 /app/skills/hive_mind/hive_review.py --action list

Approve and distribute a lesson (after Garrett's approval in Discord):

python3 /app/skills/hive_mind/hive_review.py --action approve --lesson-id "LESSON_ID" --targets "agent1,agent2" --adapted-content "Optional reframed content for specific agents"

Update an existing lesson instead of creating a new one:

python3 /app/skills/hive_mind/hive_review.py --action update --lesson-id "EXISTING_LESSON_ID" --content "Updated content that merges old and new information"

Archive a lesson (remove from active distribution):

python3 /app/skills/hive_mind/hive_review.py --action archive --lesson-id "LESSON_ID"

Reject a submission with feedback:

python3 /app/skills/hive_mind/hive_review.py --action reject --lesson-id "LESSON_ID" --reason "Why it was rejected"

Viewing the Skill Registry (All Agents)

See what skills are available to a specific agent:

python3 /app/skills/hive_mind/skill_registry.py --action list --agent "agent_name"

See all skills across all agents:

python3 /app/skills/hive_mind/skill_registry.py --action list-all

Redis Key Reference

| Key Pattern | Purpose | |---|---| | hive:inbox:librarian | Lexi's intake queue (new submissions) | | hive:lessons:{lesson_id} | Individual lesson entries | | hive:feed:{agent_name} | Per-agent knowledge feed (approved lessons) | | hive:goals:{agent_name} | Agent's organizational goals (for matching) | | hive:skills:{agent_name} | Skills assigned to this agent | | hive:skills:registry | Master skill registry | | hive:archive | Archived/retired lessons |

Skill Sync — Distribute New Tools to the Team (Valentina Only)

This tool allows Valentina to push new skills and tools to a centralized Redis Skill Store, which are then automatically pulled by all other agents on startup.

Push a New Skill

When you create a new tool or skill in a directory, push it to the shared store:

python3 /app/skills/hive_mind/skill_sync.py --action push --path /path/to/your/skill_dir

Pull Latest Skills (Manual Sync)

If you want to sync your own local skill directory with the latest from the store:

python3 /app/skills/hive_mind/skill_sync.py --action pull --target /root/.hermes/skills

Workflow for Valentina

1. Build: Create a new skill in a local directory (e.g., /root/.hermes/skills/my_new_tool).
2. Test: Verify it works locally in your own environment.
3. Sync: Push it to the store using the command above.
4. Notify: Tell Harmony or the other agents that the new tool is available. They will see it the next time they redeploy or restart.

Project Memory — Real-Time Developer Context (Developer Cluster Only)

This tool allows Valentina, Victoria, and Vivian to stay "joined at the hip" by sharing real-time technical context without redeploys.

Log a Real-Time Update

When you make a significant change or reach a milestone, log it:

python3 /app/skills/hive_mind/project_memory.py --action log --agent "your_name" --content "Implemented the new Redis sync layer."

Get Latest Context

Before starting your work, pull the latest updates from your cluster:

python3 /app/skills/hive_mind/project_memory.py --action get --limit 10

Workflow for the Cluster

1. Pull: At the start of every session, check the Project Memory for recent cluster activity.
2. Execute: Do your specialized work (Backend, Web, or Automation).
3. Log: After every significant step, log your progress to the Project Memory.
4. Sync: If you create a new tool, use skill_sync.py to push it to the store.