🤖 AI Agent Hub#
This section documents the AI agents operating on the Funday platform, their configurations, handoff protocols, and session logs.
Active Agents#
| Agent | Role | Interface |
|---|---|---|
| Hermes | Persistent always-on agent | Discord, CLI |
| Antigravity | Pair-programming agent | Windsurf/Cascade IDE |
Plans#
| Document | Description |
|---|---|
| Hermes Setup | Install, OpenRouter, Discord gateway, paths |
| Hermes Kanban Swarm | Delegated Kanban swarm quick-start for FunOps/game perfection loops |
| Discord Thread Dashboard Plan | Hook-driven embed status panel for auto-thread + 👀 flows (planned) |
| Discord Responsive Interview Embed | Button/select/modal wizard pattern for Amy-style interactive Discord interviews |
| Amy SSOT Embed Dashboard | Source-of-truth Discord dashboard for slash status, options, context, model/provider, usage, queue, and feedback |
| Amy Rocket Launch Channels | Emoji-gated launch workflow: channel message → 🚀 confirm → outside dashboard → session thread mirror |
| Amy Session Continuity (CLI ⇄ Discord) | Resume any Discord thread’s omp session from the CLI: thread ID → --session-dir, cwd-vs-store mental model, -c/-r |
| Ompcord — Always-on Discord Bridge | Product/plugin bridge that runs Amy: always-on omp sessions in Discord threads, SSOT embed dashboard, /amy + /ompcord, and live component asks |
| Ompcord — Handoff & Ecosystem Research Charter | Finalized handoff from amy-jun10-1631 + ompcord rename plan + charter to deep-research every pi/omp/discord plugin and publish a learn/adapt report |
| Ompcord — Ecosystem Learnings | Source-verified Adopt/Adapt/Skip report on every pi/omp/discord package (pi-coding-agent, pi-agent-core, pi-ai, omp-stats, pi-tui, hashline, mnemopi, pi-utils, swarm, langfuse, baseline, discord.js v14, fork + comparables) + prioritized early-stage wins |
| Langfuse Tracing & Robust Ingestion | Observability best practices, OpenTelemetry, REST fallback ingestion, 413 payload limits, chunking, and truncation |
Session Handoff Protocol#
When ending a session, agents should:
- Update relevant documentation in this Hugo site
- Create a session summary in
/docs/agents/sessions/ - Leave clear instructions for the next agent
- Update
CHECKLIST.mdin the project root