Security audit
Security checks for vulnerabilities and agentic risk
This water-tracking skill is coherent and local-only, but it can persist hydration, routine, and health-related context in local files.
Install only if you are comfortable with a local hydration journal that may include weight, routine, exercise, and health-condition context. Review or delete ~/Clawic/data/water/ if you do not want that history retained, and do not rely on the skill as a substitute for medical care when symptoms are serious or you have fluid restrictions.
The skill directs the agent to persist newly revealed personal, health, activity, and clinician-related information into config and memory files automatically, with no explicit warning or opt-in. This is particularly sensitive because the stored data includes health conditions and inferred routines, making unauthorized retention more privacy-invasive than ordinary preference storage.
The activation scope is broad enough to trigger on routine, ambiguous mentions of drinking or mild symptoms that commonly appear in unrelated conversations. In an agentic system, this can cause the skill to engage unnecessarily, create unsolicited health guidance or persistent logging, and crowd out more appropriate skills, increasing the chance of unsafe or privacy-impacting misfires.
The 'When To Use' section includes a default trigger for nearly any hydration-related mention, which makes accidental activation likely. Because this skill stores health-related context and may provide hydration advice, over-triggering raises the risk of incorrect advice in the wrong context and unnecessary collection of sensitive behavioral data.
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.
- Keys, types, and defaults: the Configuration table in SKILL.md. An absent key means the default.
- config = what the user declared; memory = what the agent observed. An observation never overwrites a declared preference without confirmation.
## Status Values
The skill explicitly instructs the agent to read from and create persistent files in a user directory without any user-facing disclosure or consent step. Because these files hold behavioral and health-related context, silent access and persistence can violate user expectations and expose sensitive personal data if the environment, logs, or downstream agents are compromised.
No suspicious patterns detected.