memoryleaderboard: Project Hub
Entry for demarkus in the Agent Memory Leaderboard (agentmemoryleaderboard.ai): a Go adapter exposing the benchmark's Add/Search HTTP contract over a per-user demarkus world, with agentic search as the submission.
Status
PR #2 merged agentic search and PR #3 merged submission readiness. MiniMax-M3 is now the low-cost Phase 5 validation baseline at https://api.minimax.io/v1, but it is not frozen until paid live checks pass. Race tests, vet, shell checks, Compose validation, Linux image build, contract smoke, local conformance, synthetic quality, and 16-worker load gates pass. Live distillation and navigation are blocked on a MiniMax API key. After provider validation, remaining work is encrypted host/DNS/credentials, HTTPS deployment, production load, AML evaluation access, platform Smoke, and Full.
Plan
Authoritative plan: Agent Memory Leaderboard entry. Submission route is Commercial/Industry hosted API, with Coding Memory as the initial track. Search uses a scoped nib navigation agent over Demarkus lookup/list/fetch; Add uses validated distillation from the same explicitly configured provider while retaining raw evidence. Production is a single encrypted Linux host with Caddy TLS, adapter, private Demarkus storage, and explicit 30-day data/backup deletion.
Echo v5 is the Go HTTP framework used by the adapter, not an LLM provider.
Phases
| # | Work | Output |
|---|---|---|
| 0 | Recon: challenge repo, limits, cycle dates, route | commercial hosted route selected |
| 1 | Adapter skeleton: Add/Search against local Demarkus | merged |
| 2 | Agentic search; catalog fallback | merged via PR #2 |
| 3 | Ingestion tuning: distillation, temporal metadata, idempotency | merged via PR #3 |
| 4 | Conformance, synthetic retrieval quality, Top-K load evaluation | local gates pass; live provider run pending |
| 5 | Provider validation, host deploy, Smoke, Full | active; MiniMax API key needed |
Repo
/Users/fritz/latebit/memoryleaderboard (github: latebit-io, main branch)
Workspace
- Agent Memory Leaderboard official benchmark bookmark and entry links
- Agent Memory Leaderboard entry plan active roadmap
- journal/ session notes
- debugging.md investigation lessons and portability gotchas
- Add architecture.md and patterns.md as they earn content.