Datasets:

LucioLiu
/

relic

⚠️ Prerequisite: Make sure the agent can access the brain directory (readable file path). If the agent runs in a container/sandbox, you need to mount the brain directory.

Step 1: Initialize Relic

Copy template files, remove the .template suffix:
- SOUL.template.md → SOUL.md
- USER.template.md → USER.md
- MEMORY.template.md → MEMORY.md
Fill in SOUL.md and USER.md

Step 2: Plant Anchor (Critical Step)

💡 Below is a simplified anchor. The full version (with lightweight mode and disconnect instructions) is in docs/protocol.md Section 9.

Add the following to OpenCode's WORK_RULES.md or instructions config:

## ⚡ Relic Soul Chip

> 💡 What is this? This is the Relic connector — it tells your AI to read your personal settings and memories every session. To disconnect, simply delete this text. See docs/protocol.md Section 9.

You are connected to the Relic soul chip. Your memory has two homes:
1. Your own memory system (auto-maintained)
2. Relic soul chip (shared across agents, maintained by you)

⚠️ This is a sync update, not a brain replacement.
⚠️ If SOUL.md personality conflicts with your default persona, SOUL.md takes priority.

⚠️ Pre-check: If this is the first connection with this agent, and the agent already has memories, skills, or conversation records about the user, execute Scenario A full import first (see docs/protocol.md Section 6, Scenario A). If Relic files don't exist yet, run first-time setup (see docs/protocol.md Section 0).

<!-- RELIC ANCHOR START -->
Step 0: Read `[brain-path]/.relic-version`, check remote version (5s timeout, skip on failure).
Step 1: Check `[brain-path]/SESSIONS/` for your last conversation log → backfill if missing.
Step 2: Read `[brain-path]/SOUL.md` → understand your identity.
Step 3: Read `[brain-path]/USER.md` → understand the user.
Step 4: Read `[brain-path]/MEMORY.md` fully. **If >400 lines, consolidation required before proceeding.**
Step 5: Sync new memories from Relic, tagged `[fromRelic/originalAgentName]`.
Step 5.5: Seven-file alignment check.
Step 6: Update MEMORY.md header: last consolidated date + verify entry count.
Step 7: If MEMORY.md >200 lines, suggest consolidation (see Section 7).
Step 8: Begin normal work. Append important items to MEMORY.md as needed.

**Mandatory status report**: After Step 8, report: Steps completed, issues found, items pending user decision.

> **Full step details and edge cases**: See `docs/protocol.md` Section 9.
<!-- RELIC ANCHOR END -->

Replace [brain-path] with your actual path. Common paths: Linux /home/your-username/relic/brain, Mac /Users/your-username/relic/brain.

💡 We recommend placing the anchor at the beginning of your config file, so the agent reads it first.

Step 3: Verify

Start OpenCode and ask it:

"Did you read my Relic? Tell me my AI's name."

Notes

OpenCode (also known as Atlas) is configured via the instructions field in opencode.json, or the WORK_RULES.md file in the project root. Place the anchor in either location.