Caura
未认领Caura (formerly MemClaw) — governed shared memory for AI agent fleets. Multi-agent, multi-tenant, MCP-native. Trust tiers, keystone policies, audit trails, knowledge graph, self-improving retrieval. Apache 2.0.
安装
claude mcp add --transport http -s user memclaw http://localhost:8000/mcp --header "X-API-Key: standalone"更多AI & ML服务器
浏览完整目录安全概况
已认领和已认证的服务器每周接受一次静态扫描，显示代码能触及的范围（外部服务、环境变量、shell 命令、代理配置目录）以及带有已知公告的依赖。认领此列表即可获得。 安全概况的工作原理
12 个工具中显示 12 个
文档中列出的工具 (12)
内容来自项目文档。服务器握手不会验证每个工具的说明或行为。
caura_doc
Structured-document CRUD, op-dispatched: write \
caura_entity_get
Look up a knowledge-graph entity by UUID.
caura_evolve
Karpathy-Loop feedback: record an outcome (success \
caura_insights
Karpathy-Loop reflection: contradictions, failures, stale, divergence, patterns, discover.
caura_keystones
Read mandatory governance rules for the current scope (tenant + fleet + agent merged). Call once per session.
caura_keystones_set
Author/remove keystone rules, op-dispatched: set \
caura_list
Non-semantic enumeration with filters, sort, cursor pagination.
caura_manage
Per-memory lifecycle, op-dispatched: read \
caura_recall
Hybrid semantic + keyword search over memories, with optional LLM-summarised brief.
caura_stats
Aggregate counts: total + breakdowns by type / agent / status. Read-only.
caura_tune
Read/update an agent's per-search profile (topk, ftsweight, freshness, blend, …).
caura_write
Single or batch (≤100) memory write; auto-enriched with type, title, summary, tags.