- SOUL.md: full agent identity, operating principles, voice - IDENTITY.md: runtime identity, hosts, boundaries - USER.md: operator context imported from hermes-soul - AGENTS.md: actual operating rules, infrastructure, quick reference - memories/curated/: 5 topics (tailscale, forgejo, agents, projects, vaultwarden) - skills/: 9 cross-harness skills imported from hermes-soul after review - docs/PLAN-CONFIGURE-PRIVATE-REPO.md: configuration plan - Validate: passes clean
20 lines
1 KiB
Markdown
20 lines
1 KiB
Markdown
# Agent Roster
|
|
|
|
| Agent | Home Host | Harness | Model | Role |
|
|
|-------|----------|---------|-------|------|
|
|
| Hermes | debby | Hermes Agent | DeepSeek | Orchestrator, primary dev agent |
|
|
| Zot | debby | Zot RPC | GLM-5.1 | Coding specialist, media workflows |
|
|
| Claude | domedog | Claude Code | Claude | Independent verification, secondary builder |
|
|
| Codex | osa | Codex CLI | GPT | FreeBSD 15 validation, ISO builds |
|
|
|
|
## Multi-agent group
|
|
|
|
"My Debby" Telegram group: @hermes_samob_bot + @zot_samob_bot. Hermes orchestrates, delegates coding/media to Zot via structured delegation.
|
|
|
|
## Secrets
|
|
|
|
All agent API keys live in Vaultwarden at vault.smilepowered.org (BW_SERVER). Bootstrap credentials (BW_CLIENTID, BW_CLIENTSECRET, BW_PASSWORD) are the only secrets in `.env` files — everything else is fetched from the vault at runtime.
|
|
|
|
## Verification
|
|
|
|
Claude (domedog) independently verifies Hermes's claims — test counts, clippy status, crate counts. Do not overstate results. Use exact `cargo test --workspace` output.
|