Security audit
Security checks for vulnerabilities and agentic risk
This title-advice skill includes broader account tools and confirmation gaps that could cause unexpected or recurring charges.
Prefer a corrected version that enforces title-only access and fixes preview charging. If using this version, configure ARI to ask before every charge, explicitly choose manual collection for one-time work, and review active schedules and competitor monitoring. Omitting --confirm does not reliably prevent spending.
scripts/ari.pyRead-only deepdive fallback can execute paid analysis
A user following the documented read-only/default workflow can incur an analysis charge. When the product is absent and --confirm was supplied, the command can additionally claim that analysis was downgraded to quote-only and no credits were charged despite reaching the paid execution path.
scripts/ari.pyRecurring paid collection can be enabled without an executable confirmation gate
These directly callable paths can start recurring credit deductions before the user has approved the recurring cost. Competitor deductions are conditional on the parent product remaining monitored.
The title-only skill contains actionable instructions for unrelated paid reports, account settings, recurring collection, and generic operations. Its final title-only restriction mitigates this but conflicts with earlier mandatory workflows, creating a credible route to out-of-scope actions and credit spending; the underlying code was not provided for verification.
This is another instance of monitoring mutations exceeding the advertised title-only scope. Creating or resuming watches can authorize recurring charges, although the required quote, explicit agreement, and accepted-price parameter are meaningful safeguards.
Creating and managing recurring watches exceeds title diagnosis and can establish continuing financial obligations. Explicit price confirmation and ownership checks substantially reduce the danger, but these capabilities remain exposed through contradictory instructions in a narrowly advertised skill.
This duplicates the out-of-scope watch-management issue: the skill exposes persistent monitoring and state-changing management commands despite its title-only contract. Required user authorization reduces the likelihood of unauthorized recurring charges, but conflicting scope instructions still create avoidable authority expansion.
Competitor association is described as automatically enabling weekly collection, which is a persistent side effect outside title diagnosis. Unlike the detailed watch flow, this instruction does not locally require disclosure and explicit acceptance of the recurring collection cost.
Recurring review collection is outside title diagnosis, and new-product collection defaults to weekly billing. The instructions require manual scheduling when recurring collection is not authorized, but relying on an agent to override a billable default creates a preventable financial-risk path.
The supplied commands expose monitoring and historical-reporting capabilities that are not declared in the title-diagnosis description. The chunk shows no title analysis or connection between these watch outputs and product-title recommendations, so these operations are not established as supporting implementation details. It does not show watch creation, advertising, or automatic publishing.
The supplied command requests a category leaderboard based on negative-review rate, a distinct analytical output not declared in the title-diagnosis purpose. No title analysis or connection to title recommendations is shown. The underlying script is unavailable, so this assessment does not infer additional API access, publishing, or other side effects.
The supplied commands expose watch-related functionality rather than product-title diagnosis or recommendations. Recurring-watch pricing and watch listing are not declared capabilities, and the chunk provides no connection establishing them as supporting steps for title diagnosis. It does not demonstrate watch creation, automatic publishing, or advertising.
The supplied commands concern monitoring features and their pricing, neither of which is declared as part of title diagnosis or recommendations. Using the ARI API is consistent with the description, but these particular capabilities are not evidently supporting implementation details for the stated task. The snippet does not show watch creation, modification, or automatic publishing.
The generic operations section describes broad audits and says this package is unrestricted, contradicting the final listing/title-only entry point. This creates scope ambiguity that could lead to unrelated analysis and spending, although the supplied document requires explicit authorization for operations charges.
The concrete concern is not merely that alerts exist: the workflow mandates fetching and proactively discussing account-wide alerts at session start. This accesses information about potentially unrelated products without a title-analysis need, although no external disclosure or automatic paid reply generation is shown.
The instruction treats an unanswered earlier quote as grounds to generate a report during a later conversation. An idle product is not evidence of current spending authorization, and server auto-confirmation can turn that inference into a real credit deduction.
run_analysis changes confirm from false to true based on a server-supplied autoConfirm value, allowing a nominal preview invocation to spend credits. This contradicts the explicit-confirmation contract and also affects callers such as deepdive; exposure depends on which commands the package whitelist permits.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- 重新发布：各市场首批上架的 1.5.0 包早于免确认修复，新注册的免费版用户仍会被要求先确认报价。
1.5.1 与当前官网下发的修复版一致，市场侧以新版本号覆盖旧包。
- 免确认只认服务端报价返回的 `autoConfirm`；采集报价扣除首次采集免费页；
非美国站首次采集可用赠送积点时不再提示「赠送积点不可用」（行为与修复后的 1.5.0 相同）。
- 命令、字段与计费规则不变，不提高最低支持版本。
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
不再停在报价等确认。费用在报告返回的 `charges` 字段透明展示（采集/分析/合计/余额前后）。
- 免确认与否**只认服务端报价返回的 `autoConfirm` 布尔**：服务端策略前 N 次付费生成不看金额上限
（新用户赠送积点按设计覆盖「3 页采集 + 一份 VOC」）；用户自设阈值优先、显式要求「每次先问」
时仍先报价；余额不足不自动。客户端不再从 `autoConfirmNote` 文案或 `autoConfirmRemaining`
自行推断——文案匹配会把 Free 小额提示里的「先确认」误判成用户要求每次询问，恰好挡住新用户。
- 采集报价同口径扣除服务端 `trialFreePages`（首次采集试用免费页），避免新用户被误报「积点不足」。
- 非美国站：服务端 `credits.gift` 策略下用户首次采集不限站点，`credits/balance` 返回 `giftAllSites`，
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
不再停在报价等确认。费用在报告返回的 `charges` 字段透明展示（采集/分析/合计/余额前后）。
- 免确认与否**只认服务端报价返回的 `autoConfirm` 布尔**：服务端策略前 N 次付费生成不看金额上限
（新用户赠送积点按设计覆盖「3 页采集 + 一份 VOC」）；用户自设阈值优先、显式要求「每次先问」
时仍先报价；余额不足不自动。客户端不再从 `autoConfirmNote` 文案或 `autoConfirmRemaining`
自行推断——文案匹配会把 Free 小额提示里的「先确认」误判成用户要求每次询问，恰好挡住新用户。
- 采集报价同口径扣除服务端 `trialFreePages`（首次采集试用免费页），避免新用户被误报「积点不足」。
- 非美国站：服务端 `credits.gift` 策略下用户首次采集不限站点，`credits/balance` 返回 `giftAllSites`，
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- 行为变更：新用户（注册并验证邮箱后）第一次要 VOC，只给 ASIN 即可（站点默认美国站），
`voc` 直接执行——自动补采集（默认 3 页）、生成报告、按服务端定价扣费并归档，
不再停在报价等确认。费用在报告返回的 `charges` 字段透明展示（采集/分析/合计/余额前后）。
- 免确认与否**只认服务端报价返回的 `autoConfirm` 布尔**：服务端策略前 N 次付费生成不看金额上限
（新用户赠送积点按设计覆盖「3 页采集 + 一份 VOC」）；用户自设阈值优先、显式要求「每次先问」
时仍先报价；余额不足不自动。客户端不再从 `autoConfirmNote` 文案或 `autoConfirmRemaining`
自行推断——文案匹配会把 Free 小额提示里的「先确认」误判成用户要求每次询问，恰好挡住新用户。
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
为 true 时赠送积点计入 `nonUsCollectCredits`、`siteNote` 为空；UK/DE/JP 卖家的第一份 VOC 同样一步直出。
- 余额不足仍在执行前拦截（402）并引导充值；定价以服务端配置为准，代码常量仅为兜底。
- 一致性：`analyze` 各类型（insight / keywords / trend / variant / compare）走同一口径，
`autoConfirm=true` 即直接生成并返回 `charges`。
- 文档：移除所有写死的单价与合计示例。采集/分析单价与页数由后台 `pricing` 配置决定，
一律以本次报价返回的 `credits` / `totalCredits` 为准，不得凭文档或记忆报数。
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- 能力层收窄：44 个专属包不再只是「声明窄」，包体 `scripts/ari.py` 同时按各自契约裁剪。
构建期用 AST 删除文档未承诺的命令、它们的解析器注册块，以及只被它们引用的辅助函数与
常量；专属包因此不再包含 `export`（任意路径写文件）、`autoconfirm`（撤销逐次确认）、
`workbench`（改写差评处理状态）、`alerts`（标记预警已读）、`competitors`、`schedule`、
`voc` / `analyze` / `charts` / `deepdive` 等母版命令。
- 新增运行期兜底闸：专属包在包内写入 `allowedCommands` 白名单，`ari.py`
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
def cmd_autoconfirm(args):
"""免确认阈值：不带参数=查看；`autoconfirm 50`=50 积点以内不问；`autoconfirm off`=每次都问；`autoconfirm default`=恢复默认。"""
value = (args.value or "").strip().lower()
if value:
if value in ("off", "ask", "0"):
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
def cmd_autoconfirm(args):
"""免确认阈值：不带参数=查看；`autoconfirm 50`=50 积点以内不问；`autoconfirm off`=每次都问；`autoconfirm default`=恢复默认。"""
value = (args.value or "").strip().lower()
if value:
if value in ("off", "ask", "0"):
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
def cmd_autoconfirm(args):
"""免确认阈值：不带参数=查看；`autoconfirm 50`=50 积点以内不问；`autoconfirm off`=每次都问；`autoconfirm default`=恢复默认。"""
value = (args.value or "").strip().lower()
if value:
if value in ("off", "ask", "0"):
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
limit = int(value)
except ValueError:
emit(error_obj("ARI_BAD_ARGUMENT", 0, "参数不对",
"用法：autoconfirm 50（50 积点以内不问）/ autoconfirm off（每次都问）/ autoconfirm default（恢复默认）"),
args.compact)
return
out = request_json("PUT", "/api/v1/user/autoconfirm", {"limit": limit})
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
limit = int(value)
except ValueError:
emit(error_obj("ARI_BAD_ARGUMENT", 0, "参数不对",
"用法：autoconfirm 50（50 积点以内不问）/ autoconfirm off（每次都问）/ autoconfirm default（恢复默认）"),
args.compact)
return
out = request_json("PUT", "/api/v1/user/autoconfirm", {"limit": limit})