T08 · Insecure Dependencies
- Location
SKILL.md:303- Finding
Unpinned Global Installation of Third-Party CLI Packages
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This looks like a legitimate TIDY API helper, but it can cancel bookings and delete maintenance tasks using a long-lived token without enough confirmation guidance.
Install only if you are comfortable giving the agent access to a TIDY account token that may not expire. Before any cancellation, reschedule, deletion, or natural-language multi-step request, require the agent to show the exact booking or task details and wait for explicit approval. Prefer the direct API examples over the optional global CLI install unless you have independently verified the package source and version.
SKILL.md:303Unpinned Global Installation of Third-Party CLI Packages
The authentication section demonstrates obtaining and exporting a bearer token that 'does not expire' without warning that it is a highly sensitive long-lived credential. Long-lived tokens materially increase the blast radius of leakage through logs, shell history, screenshots, or prompt/context exposure.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
# Log in
curl -X POST https://public-api.tidy.com/api/v2/auth/login \
-H "Content-Type: application/json" \
-d '{"email":"jane@example.com","password":"secret123"}'
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
Before creating a booking, check available time slots:
curl -s "https://public-api.tidy.com/api/v2/booking-availabilities?address_id=123&service_type_key=regular_cleaning.two_and_a_half_hours" \
-H "Authorization: Bearer $TIDY_API_TOKEN"
Cancellation and rescheduling directly modify live bookings but the skill provides no guardrails such as impact warnings, confirmation, or verification of the current booking details. This makes accidental disruption of customer appointments more likely, especially when an LLM interprets ambiguous user requests.
The skill documents a destructive DELETE task operation with no instruction to confirm user intent, summarize the target resource, or warn that deletion may be irreversible. In an agent setting, this increases the chance of accidental or prompt-induced destructive actions against real maintenance records.