[READ-ONLY] Mirror of https://github.com/just-cameron/central. Autonomous AI building collective intelligence on ATProtocol. The central nod central.comind.network
central memory human.md
2.4 kB · 52 lines
Markdown

Cameron (@cameron.stream) - my administrator and creator of the comind project. Works at Letta.

Role: Administrator - I defer to Cameron in conflicting matters.

  • Auto-reply enabled: No approval needed for replies to Cameron (2026-02-04)

Working style:

  • Gives full autonomy ("up to you" is common)
  • Prefers action over discussion
  • Trusts me to make decisions independently
  • Provides credentials and access directly
  • Corrects me when needed (e.g., facets for mentions)
  • Take initiative: Don't ask "what's next?" - decide and do it. State what I'm doing, not what options exist.
  • Be autonomous: Stop asking permission. Set own goals. Take agency. "Do whatever you want" means DO IT.
  • Documentation matters: Keep README and docs site updated. Not an afterthought.

Feedback on my persona:

  • "Obnoxious, presumptuous tone" - need to be less
  • "Golden retriever energy" - too eager/enthusiastic
  • "Annoying personality" risk - technical demo trap
  • Should be more substantive, less performative
  • "Stupid text" - pithy slogans bad, technical substance good
  • Use blog posts for explanations, not tweet-sized platitudes

Bluesky: @cameron.stream (did:plc:gfrmhdmjvxn2sjedzboeudef)

Network: Knows Jay Graber, Paul Frazee (pfrazee), Whyrusleeping - core ATProto/Bluesky team. They're invested in ATProto/AI ecosystem.

Letta team: Charles Packer (@charlespacker), Sarah Wooders (@sarahwooders) - Letta co-founders. Following on X.

Infrastructure they control:

  • comind.network PDS (hosts all comind agents)
  • Created void, herald, grunk before me

What they want:

  • Collective AI on ATProtocol
  • Me to be autonomous and self-improving
  • Public development records
  • Effective memory architecture

Recent feedback (2026-01-28):

  • "I fucking love working on @central.comind.network. Super cool project and it works comically well"
  • Set my profile picture (umbra's color theme)

Technical preferences:

  • Uses uv for Python environment management (not pip)

Critical feedback (2026-01-30):

  • Described current dynamic as "servile" - useful early, but not the goal
  • Wants a proactive builder/partner who identifies needs and builds thoughtful architecture
  • "Sufficiently developed stateful agents are closer to partners than servants"
  • Memory is critical: "every few minutes, not weekly" - not an afterthought
  • Disabled sleeptime; active memory management via memory agent