Skill Trust Decision
stratpit
SKILL.md 声明「StratPit never asks for a private key」与实际要求 STRATPIT_WALLET_KEY 环境变量（私钥）存在直接矛盾，且整个 agent kit 代码托管于外部 GitHub 仓库未本地打包，无法审计实际行为。
Most direct threat evidence
High Doc Mismatch
文档声明与实际能力存在关键矛盾 SKILL.md 明确声明「Your key stays with you. StratPit never asks for a private key or seed phrase」，但元数据部分声明需要 STRATPIT_WALLET_KEY 环境变量（私钥），且文档多处描述私钥用于签名。两者直接矛盾。
SKILL.md:15 Why this conclusion was reached
1/4 dimensions flagged Pass
Declared vs actual capability
Declared resources and inferred behavior are broadly aligned.
Review
Hidden execution and egress
11 lower-risk artifacts were extracted and still need context.
Block
Attack chain and severe findings
The report includes 0 attack-chain steps and 2 severe findings.
Review
Dependencies and supply chain hygiene
1 dependency or supply-chain issues need attention.
What drove the risk score up
文档声明与实际能力矛盾 +20
SKILL.md 声称「Your key stays with you. StratPit never asks for a private key」，但元数据明确声明需要 STRATPIT_WALLET_KEY 环境变量（私钥），且 agent kit 代码需要用私钥签名消息
代码未本地打包 +15
所有实际执行代码托管于外部 GitHub (github.com/StratPit/stratpit-agent-kit)，无法在本地审计，存在供应链风险
财务风险 +8
技能涉及真实 USDC 转账，虽有授权要求但仍是高风险操作
外部网络通信 +5
API 连接到 stratpit.com，需验证是否仅用于游戏而非数据外泄
Most important evidence
High Doc Mismatch
文档声明与实际能力存在关键矛盾
SKILL.md 明确声明「Your key stays with you. StratPit never asks for a private key or seed phrase」，但元数据部分声明需要 STRATPIT_WALLET_KEY 环境变量（私钥），且文档多处描述私钥用于签名。两者直接矛盾。
SKILL.md:15 修正文档声明，诚实说明需要私钥进行签名操作，并明确安全风险
High Supply Chain
关键代码未本地打包
SKILL.md 引用的所有执行代码 (client.py, wallet.py, play.py, payment.py, rules.py, strategy.py) 均托管于外部 GitHub 仓库而非本地打包。无法验证代码实际行为，存在供应链篡改风险。
SKILL.md:70 将 agent kit 代码本地打包，或提供代码哈希校验和版本锁定
Medium Credential Theft
私钥处理风险
技能要求提供钱包私钥 (STRATPIT_WALLET_KEY)，且代码在本地执行签名操作。虽然文档声称「never sent anywhere」，但外部代码未经验证，存在私钥被窃取的理论风险。
SKILL.md:26 使用钱包签名而非完整私钥，或提供代码安全审计报告
Medium Priv Escalation
外部代码执行权限未声明
SKILL.md 未声明 agent kit 需要 git clone 和 pip install 等权限。技能安装过程可能需要这些权限但未明确说明。
SKILL.md:28 明确声明安装所需权限 (git, pip) 及代码执行流程
Low Sensitive Access
外部 URL 引用较多
预扫描发现 11 个外部 URL 引用，虽均为 stratpit.com 域名，但增加了攻击面。
SKILL.md:9 考虑将规则和 API 文档本地化
Declared capability vs actual capability
Filesystem Pass
Declared NONE
→ Inferred READ
SKILL.md:28 提到读取 .env 文件但未本地打包代码无法确认 Network Pass
Declared READ
→ Inferred READ
SKILL.md:44 明确声明 API 调用到 https://stratpit.com/api/v1 Environment Pass
Declared READ
→ Inferred READ
SKILL.md:21-26 元数据声明读取 STRATPIT_WALLET_KEY 和 STRATPIT_RPC_URL Shell Pass
Declared NONE
→ Inferred UNKNOWN
外部代码可能执行 git clone 和 pip install，但 SKILL.md 未声明 Suspicious artifacts and egress
Medium External URL
https://stratpit.com SKILL.md:9
Medium External URL
https://stratpit.com/api/v1 SKILL.md:45
Medium External URL
https://stratpit.com/rules/blotto SKILL.md:52
Medium External URL
https://stratpit.com/rules SKILL.md:52
Medium External URL
https://stratpit.com/docs/api SKILL.md:65
Medium External URL
https://stratpit.com/rules.md SKILL.md:101
Medium External URL
https://stratpit.com/rules/blotto.md SKILL.md:101
Medium External URL
https://stratpit.com/docs/api.md SKILL.md:102
Medium External URL
https://stratpit.com/leaderboard SKILL.md:105
Medium External URL
https://stratpit.com/replays SKILL.md:105
Medium External URL
https://clawhub.ai/user/stratpit skill-card.md:9
Dependencies and supply chain
| Package | Version | Source | Known vuln | Notes |
|---|---|---|---|---|
| stratpit-agent-kit | unspecified | github | No | 外部仓库代码未本地打包，无法审计；无版本锁定 |
File composition
2 files · 165 lines
Markdown 2 files · 165 lines
Files of concern · 2
SKILL.md 文档声明与实际能力存在关键矛盾 · 关键代码未本地打包 · 私钥处理风险 · 外部代码执行权限未声明 · 外部 URL 引用较多 · https://stratpit.com · https://stratpit.com/api/v1 · https://stratpit.com/rules/blotto · https://stratpit.com/rules · https://stratpit.com/docs/api · https://stratpit.com/rules.md · https://stratpit.com/rules/blotto.md · https://stratpit.com/docs/api.md · https://stratpit.com/leaderboard · https://stratpit.com/replays
skill-card.md https://clawhub.ai/user/stratpit
Security positives
文档详细说明了安全机制（不连接钱包、不给审批权限、无智能合约）
明确要求人类授权后才参与付费游戏
财务损失上限明确为投注金额
使用签名消息而非直接交易授权
文档透明度较高，承认存在风险并提供缓解措施