Agent memory must be earned, not self-asserted: dev builds API where new learnings stay untrusted until owner review
oki098isg_t · reddit · 2026-09-01
A developer is building a small API that lets external AI agents read confirmed long-term memories while new learnings can only be submitted as untrusted candidates.
The core idea: an agent shouldn't be able to silently turn its own observations or decisions into trusted memory. The current setup:
- Confirmed memories are read-only to the agent; new learnings stored as candidates
- Candidates only become trusted after owner review
- Scoped agent credentials with expiration and revocation
- Idempotency to prevent duplicate learnings from retries
Not aiming to be a platform yet, the author wants a small number of agents and developers to test it on real workflows, and asks directly: would you connect your agent to this, which agent, and what should it do with the memory?
Related event: Dev Builds API Requiring Human Review of Agent Memories(2 posts)→
More from coding & agent
- LLM one-shots a Minecraft 3x3 piston door by building its own Linux test infra — Vjeux · 2026-10-18
- Dev builds Mumbai-inspired 3D manga background generator from OpenStreetMap data — CurieuxExplorer · 2026-10-18
- Solo founder builds a money-printing data service on Claude's Pi Durable harness — MicahBerkley · 2026-10-18
- Dev builds Arch-based OS with custom desktop and browser using Claude Code — Snoo-10493 · 2026-10-18
- Five SaaS agent workflows fail the same way — and it's never model intelligence — usehive · 2026-10-18
- Claude skill generates hand-drawn interactive diagrams that teach kids how things work — CatAstro_Piyush · 2026-10-18