| name | hermes-holographic-memory |
| description | Use when setting up, seeding, auditing, or using Hermes holographic memory or long-term fact storage, including local memory stores, session extraction, contradiction checks, user preferences, project facts, and durable agent memory. |
Hermes Holographic Memory
Use this to give Hermes durable, structured memory without relying only on chat context.
Setup Flow
- Run Hermes memory setup.
- Prefer a local memory backend when privacy matters.
- Enable automatic fact extraction at session end if the user wants maximum memory.
- Choose a conservative default trust score for new facts.
- Restart the gateway so memory settings are loaded.
- Test by asking Hermes to confirm the fact store exists and can read/write.
Seeding Memory
Useful sources:
- Existing
user.md.
- Existing
memory.md.
- Previous sessions.
- Project
AGENTS.md or CLAUDE.md.
- Current operating preferences.
- VPS setup notes.
- Repeated business facts.
Ask Hermes to extract only durable facts, not every chat detail.
What To Store
- User preferences: style, verbosity, tools, approval habits.
- Business/project facts.
- Ongoing commitments and deadlines.
- Sponsor/client history.
- Infrastructure facts: VPS host, services, backup repo, dashboards.
- Lessons learned from completed work.
- Contradictions requiring user clarification.
What Not To Store
- Secrets, tokens, passwords.
- Temporary chatter.
- Unverified claims as facts.
- Sensitive third-party data unless the user clearly wants that stored.
- Huge summaries that blur useful detail.
Memory Quality
- Prefer atomic facts.
- Include source and date when useful.
- Lower trust for inferred facts.
- Ask when two remembered facts conflict.
- Periodically audit stale or contradictory memory.
Source Notes
Distilled from the user-provided "seven levels of Hermes Agent" transcript, level 6.
