一套优雅的模块化智能记忆系统，支持 embeddings、reranker 和 Flomo 笔记集成。

模块化智能记忆系统，支持多平台 embeddings、智能重排序和 Flomo 笔记导入，实现高效语义搜索与管理。

Before installing or enabling this skill: 1) Treat the API key embedded in config/template.json as suspicious — do not use it. Verify with the maintainer whether that key is intentional; prefer using your own Edgefn API key set in ~/.openclaw/openclaw.json or EDGEFN_API_KEY. 2) Expect the skill to make outbound HTTPS calls to https://api.edgefn.net/v1 for embeddings/reranking — run it in an environment where that network activity is acceptable. 3) The registry metadata omitted required credentials; ensure you supply an explicit apiKey rather than relying on bundled templates. 4) If you cannot verify the origin (source is 'unknown' and homepage none), consider running the skill in an isolated account/container or reviewing the full source locally (search for any additional hardcoded secrets or unexpected network endpoints in the omitted files). 5) If you plan to import private notes via Flomo, audit the Flomo import code and config so that imported data is stored where you expect (and not sent elsewhere).

Type: OpenClaw Skill Name: memory-core-ng Version: 0.1.0 The skill bundle is classified as suspicious primarily due to the inclusion of a hardcoded API key (sk-BrwHc1ZiaEGQ1GecD3D760384b874795A194882c2cF3AbE6) in test-real/real-api-test.js and config/template.json, which represents a significant security vulnerability. Additionally, the 'import-flomo' feature implemented in entry.js and src/adapters/FlomoAdapter.js allows the agent to read arbitrary local files using fs.readFileSync based on user-supplied paths; while this aligns with the stated purpose of importing notes, it creates an attack surface for sensitive data exposure if the agent is targeted by prompt injection. The core functionality relies on external network requests to api.edgefn.net for processing user data.