Dream

A proactive memory distillation skill that maintains MEMORY.md. Trigger scenarios: (1) Scheduled distillation at 03:30 daily (runs when idle, deferred to 06:...

This skill appears coherent and implements what it claims: local distillation, archiving, and re-emergence tracking. Before installing: (1) verify the skill's source (SKILL.md references https://github.com/teman2050/dream-skill but registry metadata had 'source unknown'); (2) back up your existing MEMORY.md and any important workspace files; (3) review dream-tools.sh yourself (it runs locally and will modify files under your OpenClaw workspace and the configured DREAM_VAULT_PATH); (4) ensure jq and the OpenClaw CLI are installed and available on PATH; (5) if you want to limit automatic writes, avoid enabling the schedule or only allow manual invocation. If you are unsure about the origin or don't want a skill to autonomously modify your memory files, treat this as a manual tool instead of enabling scheduled/autonomous runs.

Type: OpenClaw Skill Name: memory-file-manager-teman Version: 0.2.2 The skill bundle is classified as suspicious due to several vulnerabilities in `dream-tools.sh` that could be exploited via prompt injection against the AI agent. Specifically, `cmd_atomic_write` has a path traversal risk, `cmd_ledger_append` is vulnerable to markdown/JSON injection, `cmd_check_reemergence` is susceptible to regex injection, and `cmd_status` has a potential Python code execution risk. The `Skill.md` instructions for the AI to 'write directly' and 'forget' content based on user input create a significant prompt injection surface, allowing an attacker to potentially trigger these vulnerabilities, leading to unintended file modifications, data corruption, or denial of service. There is no clear evidence of intentional malicious behavior like data exfiltration or backdoors.