47/100 (Medium)

OpenClaw： suspicious
VirusTotal： suspicious
StaticScan： clean

OpenClaw： suspicious

The skill's stated purpose (agent self‑improvement using local logs and cron scripts) mostly matches what it reads and writes, but its instructions allow broad file reads and committing workspace chan...

[内容已截断]

VirusTotal： suspicious VT 报告

静态扫描： clean

No suspicious patterns detected.

README 未提供

{
    "latestVersion": {
        "_creationTime": 1774501754129,
        "_id": "k97cm6thcrcw07m040c6jan23d83m3ce",
        "changelog": "v1.1.1: Added external memory rule — Success once ≠ Learned.\n\nCore problem: AI agents restart fresh each session. Even after successfully completing a task (e.g. reading Apple Calendar, fetching data), the agent may fumble the same task next time if the verified path is not written down.\n\nNew rule added to Core Rules (rule #7):\nA task is only truly learned after the verified path is written into external memory — TOOLS.md, improvement_log.md, or long-term MEMORY.md.\n\nTrigger: externalize any workflow that is (1) already verified, (2) likely to repeat, (3) has easy-to-forget details.\n\nThree-layer external memory structure:\n- TOOLS.md: verified tool paths, entry commands, do-not-detour conclusions\n- improvement_log.md: pit-fall records, weekly improvement notes\n- MEMORY.md: long-term rules and preferences explicitly requested by the user\n\nbehavior-learning.md updated: promotion rules now include — verified repeatable workflows promote immediately to TOOLS.md (no need to wait for 3 occurrences); user-explicit long-term preferences also sync to MEMORY.md.\n\nNew definition of done: task success + verified path externalized = truly learned.",
        "changelogSource": "user",
        "createdAt": 1774501754129,
        "parsed": {
            "clawdis": {
                "author": "paoloxiamn"
            }
        },
        "version": "1.1.1"
    },
    "owner": {
        "_creationTime": 0,
        "_id": "s170ce23b0frjm0wkfzwpdh1qs83g1dj",
        "displayName": "PaoloXiaMN",
        "handle": "paoloxiamn",
        "image": "https:\/\/avatars.githubusercontent.com\/u\/250343763?v=4",
        "kind": "user",
        "linkedUserId": "kn79tq208tpsp8ymzdv1p1yhex82t0ay"
    },
    "ownerHandle": "paoloxiamn",
    "skill": {
        "_creationTime": 1774346249969,
        "_id": "kd7fn1kt9a6rwcddcaax3bdwch83ggrf",
        "badges": [],
        "createdAt": 1774346249969,
        "displayName": "LX Agent Optimizer",
        "latestVersionId": "k97cm6thcrcw07m040c6jan23d83m3ce",
        "ownerPublisherId": "s170ce23b0frjm0wkfzwpdh1qs83g1dj",
        "ownerUserId": "kn79tq208tpsp8ymzdv1p1yhex82t0ay",
        "slug": "lx-agent-optimizer",
        "stats": {
            "comments": 0,
            "downloads": 63,
            "installsAllTime": 0,
            "installsCurrent": 0,
            "stars": 1,
            "versions": 3
        },
        "summary": "A battle-tested agent self-improvement and optimization system built from real-world usage. Combines behavior learning, proactive patterns, cron discipline,...",
        "tags": {
            "latest": "k97cm6thcrcw07m040c6jan23d83m3ce"
        },
        "updatedAt": 1774501804849
    }
}