Openclawdy

Memory infrastructure for AI agents. Persistent storage, semantic recall, reputation tracking, cross-agent pools, and time-travel snapshots. Wallet-based aut...

This skill appears to be a third‑party memory service (openclawdy.xyz) and is internally consistent with that purpose, but proceed cautiously: - Verify operator/trust: confirm who runs openclawdy.xyz and review their privacy/legal terms before sending any sensitive data. - Signing vs keys: the service requires signed headers. Confirm how your agent will produce signatures — ensure private keys remain local and the agent only sends signatures (not private keys). The registry should document the signer integration; ask the author if it does not. - Data exfiltration: the API supports full vault export and storage of arbitrary content. Do not grant this skill to agents that can access secrets, PII, or other sensitive files unless you accept that those may be stored off‑site. - Supply‑chain risk: the install instructions point to a remote SKILL.md URL; prefer installing from the registry snapshot or a vetted source rather than fetching a live remote file you don't control. - Least privilege: only enable/use this skill for agents that need long‑term memory and explicitly consent to third‑party storage. If possible, run a self‑hosted or audited alternative. If you want a higher assurance decision, ask the skill owner for: (1) documentation of the signing integration (how signatures are made/verified and whether any SDK/plugin is required), (2) operator identity and data processing agreement, and (3) a static, versioned SKILL.md hosted in the registry rather than only on an external URL.

Type: OpenClaw Skill Name: openclawdy Version: 1.1.0 The OpenClawdy skill bundle defines an API for an AI agent memory service. All interactions are via HTTPS calls to the documented `https://openclawdy.xyz` endpoint. The `SKILL.md` explicitly states that no environment variables or external binaries are required, significantly reducing the attack surface for local execution or credential theft. Authentication is wallet-based signing, designed to prevent private key exposure. There are no prompt injection attempts, hidden commands, data exfiltration instructions, or other indicators of malicious intent within the provided files. The functionalities, while powerful (e.g., `memory_export`, `memory_clear`), are clearly documented features of a memory management system, and their misuse would stem from agent instructions rather than the skill's inherent design.