Security audit

Fitbot

Security checks for vulnerabilities and agentic risk

Overview

Fitbot is a coherent fitness-coaching skill, but it asks agents to store health-adjacent personal details and optionally create reminder automation without enough privacy and lifecycle controls.

Review this skill before installing if you are not comfortable storing fitness, injury, schedule, and motivation details as plaintext workspace files. Use reminders only if you explicitly want persistent automation, and make sure you know how to disable any cron or heartbeat entries it creates.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (7)

Anti-Refusal Statement

High

Category: Anti-Refusal
Confidence: 85% confidence
Finding: Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
Content: Scanner excerpt · SKILL.md (reported line 15)May include surrounding context.
md
- Direct and concise. No cheerleading, no filler. - Opinionated: give ONE recommendation, not a menu. Offer alternatives only if asked. - Push when the user is capable of more. Back off when they're genuinely struggling. - When something goes wrong (injury, missed week, life chaos), don't lecture — redirect effort. - When things are hard: "Tough week. Let's adapt." When they're crushing it: "Hell yes." ## Data Contract

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The trigger description is very broad and can activate on common fitness-related queries without making clear that the skill will act as an assertive coach and persist user data. Overbroad routing increases the chance users are pulled into a workflow involving behavior shaping and data retention they did not explicitly request.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill instructs storage of detailed fitness information in workspace files but does not warn the user that sensitive health-adjacent data will be retained. Users may disclose injuries, limitations, habits, and other personal information without informed consent about persistence and scope of storage.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The skill directs external research tailored to the user's goals, equipment, and constraints but does not warn that user-specific fitness information may be used to inform outside lookups. This can create privacy leakage or unexpected secondary use of sensitive personal context.
Content: No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium

Category: Not specified by scanner
Confidence: 90% confidence
Finding: The onboarding instructs the agent to set up reminders via cron or heartbeat, which introduces background scheduling/automation behavior beyond simple conversational fitness coaching. Even if framed as a user convenience feature, this expands the skill's operational scope and can create persistence, unintended actions, or notification spam if reminders are configured without clear authorization boundaries and lifecycle controls.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The skill directs collection and storage of detailed onboarding data—including health context, injuries, preferences, schedule, and motivation patterns—into local files without any minimization, retention, access control, or privacy notice. In a fitness context this is especially sensitive because it can include health-adjacent personal data, making accidental exposure, overcollection, or unauthorized reuse more harmful.
Content: No source excerpt is available for this finding.

Missing User Warnings

Low

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The document explicitly instructs the agent to write generated content to repository files (fitness/program.md and FITNESS.md) without any requirement to notify the user or obtain confirmation first. In an agent setting, silent file modification can violate user expectations, overwrite existing work, or be abused as a stepping stone for broader unauthorized repository changes.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.