Shows all active Telegram threads, background loops, and operations currently running. Tracks that exceed their expected max duration are marked as stale.
Loading…
Loading…
Loading…
You can spawn a fresh instance of the TrueSight DAO Autopilot on any Linux server. Each instance gets its own identity, context, and transcript repo.
git clone https://github.com/TrueSightDAO/truesight_autopilot.gitpip install -r requirements.txtcp .env.example .env and fill in your keys (see below)https://github.com/TrueSightDAO/agentic_ai_context - replace the content with your own runbooks and protocolsmy-autopilot-transcript) - the autopilot writes session history hereAGENTIC_CONTEXT_REPO and TRANSCRIPT_REPO at your forkspython3 app/main.py - the vault page at /vault/status will confirm it is alive| Credential | Env Var | Why |
|---|---|---|
| LLM API Key | DEEPSEEK_API_KEY or LITELLM_API_KEY | Required to reason and respond. Without it, the autopilot cannot function. |
| GitHub PAT | GITHUB_PAT | Required to read repos, open PRs, and submit contributions. |
| Telegram Bot Token | TELEGRAM_BOT_TOKEN | Required to receive and send messages. Get one from @BotFather. |
A governor can already say:
"Sophia, spawn a new instance for Bilal" (clone) or
"Sophia, fork the ecosystem for Liz" (new DAO).
There are two paths:
Same DAO, new operator. New Sophia shares the existing Edgar API and ledger.
New DAO, new rules. New Sophia + new Edgar + new DB + blank context.
Try it: Say "Sophia, spawn a new instance for [name]" or "Sophia, fork the ecosystem for [name]".
I'll follow the
Self-Replication SOP.
The new instance's /vault/status page will show its own runtime config.
Each instance is content-addressable - its behavior is determined by its context repo, not hardcoded logic. Two instances with different context repos will behave completely differently, even running the same code. The transcript repo is per-instance memory - it starts empty and fills as the instance operates.
Loading…