Security audit
Security checks for vulnerabilities and agentic risk
This skill can spend your ARI credits without fresh approval, including through a command documented as quote-only.
Prefer a release that fixes quote-only execution. If using this version, set and verify 'ask every time' with autoconfirm off, use quote for free pricing, and explicitly approve recurring collection costs before enabling schedules or binding competitors.
scripts/ari.pyQuote-only deepdive can silently execute paid analysis
The CLI submits paid VOC analysis without the explicit confirmation promised for deepdive. The same promotion defeats the intended quote-only fallback when no main-product subscription is found; the resulting product note can incorrectly claim that no credits were deducted.
The broader feature set alone is not a vulnerability, but it includes consequential spending behavior beyond the advertised pre-launch research task. In particular, the competitor-comparison workflow recommends binding competitors, which it says starts weekly collection, without an explicit recurring-cost confirmation at that step; the separate watch workflow has stronger consent requirements.
Monitoring competitors can support product research, but persistent watch lifecycle management and recurring price-accepted refreshes materially extend the declared pre-launch review-validation purpose. The supplied chunk does not show review analysis or demand validation; it shows scheduled monitoring operations. Because this is a command listing rather than implementation code, it establishes the exposed capabilities but does not establish that any command was executed or any charge incurred.
The declared purpose describes pre-launch review research, but the supplied commands manage ongoing weekly monitoring and its associated per-refresh pricing. These are substantive recurring operations not disclosed in the description, rather than merely implementation details of review analysis. The supplied chunk does not itself show the declared review-analysis workflow.
Although competitor monitoring can inform product research, the declared description is limited to pre-launch validation through review analysis. It does not disclose persistent scheduled monitoring or acceptance of recurring refresh charges. These are substantive additional capabilities, not merely implementation details. The chunk contains CLI examples rather than the underlying implementation, so it establishes the exposed command capabilities but does not prove that any watch was created or charge incurred.
Review collection supports the declared research purpose, but configuring persistent, potentially charge-incurring monitoring is a materially additional capability not disclosed in the description. The supplied chunk shows scheduling commands rather than research analysis; the underlying scheduler implementation and actual billing behavior are not available for verification.
The supplied commands concern watch configuration/status and monitoring costs rather than review-based product validation. This is an additional undeclared capability. The snippet does not show watch creation, actual monitoring execution, or the underlying API implementation, so those behaviors cannot be inferred.
The skill instructs the agent to revive an unanswered quote in a later conversation and proceed with generation when automatic confirmation applies. An idle product records prior interest, not current authorization, so this can consume credits for work the user never approved or no longer wants.
run_analysis replaces an absent local confirmation with a server-supplied autoConfirm value, then submits a credit-consuming analysis. This contradicts the explicit-confirmation contract and can make analyze or deepdive spend credits when invoked expecting only a quote; deepdive's intended quote-only fallback is also affected.
This entry preserves the automatic-confirmation behavior described in v1.5.0, where a server-generated flag permits paid operations without an initial price confirmation. In this context, the flag represents a service policy, not necessarily the user's authorization to spend credits; this is the same underlying consent issue as the later findings, not a separate exploit.
- 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。
The documented default allows the first N paid generations without an amount limit, triggered by a general report request rather than approval of its price. Honoring explicit 'ask every time' preferences and checking the balance reduce exposure, but neither establishes affirmative spending consent; reporting charges afterward cannot prevent unwanted credit consumption.
- 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。
Using a structured boolean instead of interpreting explanatory text is sensible, but the surrounding policy makes that boolean sufficient to authorize uncapped introductory paid operations. The vulnerability is the absence of a default user-approved spending boundary, not the choice of a boolean field; this duplicates the consent issue at line 32.
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`，
This is a duplicate match on the same automatic-spending policy, not an additional vulnerability. Although the policy respects user-defined thresholds when present, its default permits paid execution based on a service-controlled flag without demonstrating that the user approved a price or budget.
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`，
This entry extends the same server-flag-based automatic charging to multiple analysis types, increasing the operations that can consume credits without a transaction-specific confirmation. A balance check prevents spending unavailable credits, but does not prevent depletion of an existing balance; the supplied changelog does not establish an independent user-approved default budget.
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})
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.
return
out = request_json("PUT", "/api/v1/user/autoconfirm", {"limit": limit})
else:
out = request_json("GET", "/api/v1/user/autoconfirm")
if ok(out) and isinstance(data_of(out), dict):
d = data_of(out)
d["note"] = AUTOCONFIRM_MODE_NOTE.get(d.get("mode"), AUTOCONFIRM_MODE_NOTE["ask"]).format(
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.
p = sub.add_parser("check", parents=[common], help="验证 Key、账户和积点余额")
p.set_defaults(fn=cmd_check)
p = sub.add_parser("autoconfirm", parents=[common],
help="免确认阈值：autoconfirm 50 / autoconfirm off / autoconfirm default；不带参数查看")
p.add_argument("value", nargs="?", default="")
p.set_defaults(fn=cmd_autoconfirm)
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.
p.set_defaults(fn=cmd_check)
p = sub.add_parser("autoconfirm", parents=[common],
help="免确认阈值：autoconfirm 50 / autoconfirm off / autoconfirm default；不带参数查看")
p.add_argument("value", nargs="?", default="")
p.set_defaults(fn=cmd_autoconfirm)
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.
p.set_defaults(fn=cmd_check)
p = sub.add_parser("autoconfirm", parents=[common],
help="免确认阈值：autoconfirm 50 / autoconfirm off / autoconfirm default；不带参数查看")
p.add_argument("value", nargs="?", default="")
p.set_defaults(fn=cmd_autoconfirm)
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.
p.set_defaults(fn=cmd_check)
p = sub.add_parser("autoconfirm", parents=[common],
help="免确认阈值：autoconfirm 50 / autoconfirm off / autoconfirm default；不带参数查看")
p.add_argument("value", nargs="?", default="")
p.set_defaults(fn=cmd_autoconfirm)