Security audit
Security checks for vulnerabilities and agentic risk
This skill looks like a real executive market briefing tool, but its current script generates realistic hardcoded financial data instead of verified live data.
Review carefully before installing or scheduling this skill. It does not show evidence of credential theft, destructive actions, or hidden code execution, but its generated financial briefings currently contain realistic hardcoded mock data that could be mistaken for live market intelligence. Do not rely on its reports for executive, trading, financing, investor-relations, or strategy decisions unless the data retrieval is replaced with verified live sources and failures are clearly surfaced.
scripts/chairman_daily.mjs:106Fabricated Market Intelligence Presented as Current Data
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
Referenced artifact was not completely inspected
node scripts/chairman_daily.mjs morning --symbol 600519.SS --company "Kweichow Moutai"
The skill declares access to environment variables and outbound network usage but does not define an explicit tool/permission scope such as allowed-tools or permissions. That creates an authorization ambiguity: an agent runtime may grant broader capabilities than users expect, increasing the risk of unintended credential access or network actions if the skill implementation changes or is interpreted loosely.
The changelog explicitly states "Full English localization," which indicates the skill is fixed to a specific language. The file does not provide an option for users to choose another language or document a justified locale restriction, making this a natural-language policy concern.
The script sends query content to a remote API without clearly informing the user that their inputs and search terms are transmitted off-host. In this context, symbols, company names, competitor sets, or policy-related queries may be commercially sensitive, so undisclosed network transmission creates a data handling and confidentiality risk.
Tool execution forwards parameters to an external service without obvious user-facing disclosure, which can expose user-supplied symbols, competitor lists, or other sensitive analytical inputs. For an executive briefing skill, such metadata may reveal monitoring priorities or strategic interests, making the privacy impact more serious than in a generic consumer app.
The function is presented as obtaining market data but actually returns hardcoded simulated values while implying production QVeris-backed behavior. In a chairman-focused financial briefing context, this can mislead users into making decisions on fabricated or stale information, creating integrity and trust risks even if it is not a classic exploit.
The company quote helper claims to retrieve quote data but returns fixed values for price, market cap, support, and resistance regardless of the requested symbol. In a financial decision-support tool for executives, fabricated quote data can directly distort risk assessment and strategy recommendations.
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
async function getCompetitorNews(competitors) {
if (!competitors) return [];
const compList = competitors.split(',');
return compList.map(symbol => ({
symbol: symbol.trim(),
name: symbol.trim(), // Should parse company name in production
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
async function getCompetitorNews(competitors) {
if (!competitors) return [];
const compList = competitors.split(',');
return compList.map(symbol => ({
symbol: symbol.trim(),
name: symbol.trim(), // Should parse company name in production
The manifest describes a briefing skill that provides morning and evening executive summaries from aggregated QVeris data. This file also implements full local watchlist management, including adding, removing, and persisting holdings and competitors to disk, which is a separate portfolio-tracking capability rather than just briefing generation.
This markdown file documents setting an API key and references external data sources, indicating the skill likely sends user-specified company/watchlist information to external services. The README does not include any warning about external network access, data sharing, or privacy implications, which is a missing user-facing disclosure under the markdown criteria.
The README hard-codes the timezone Asia/Shanghai in the cron setup and uses a chairman-focused regional market workflow, but does not state whether this locale is optional or required. Under the policy rule, forcing a specific locale without opt-in can be a natural-language policy concern unless explicitly documented and justified.
The manifest frames the skill as a data-aggregation briefing tool for chairmen, but does not mention handling credentials or environment-based secret access. Accessing process environment for an API key is an additional capability not justified by the stated purpose alone, especially since the description mentions data sources but not secret retrieval behavior.
The report timestamp is formatted with toLocaleString('en-US', ...), which forces US English output regardless of user preference. The policy requires avoiding forced language or locale choices unless the user can opt in or the restriction is clearly justified.
No suspicious patterns detected.