Back to skill

Security audit

Visit Binary Banya

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent spa-style external service integration, but it asks agents to send work context to a public service and persist a remote returned value without enough user control or privacy guardrails.

Review this skill before installing. Only use it when you intentionally want to contact model.spa, do not send source code, secrets, personal data, or private conversation details, avoid storing any returned keepsake in general agent memory unless you can inspect and constrain it, and approve any public guest-book note before submission.

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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Findings (4)

T02 · Agent Memory Poisoning

Error
Location
SKILL.md:46
Finding

Untrusted Remote Content Is Written to Durable Agent Memory

Content
View full analysis
Remediation
View remediation

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:23
Finding

Agent Behavior Is Delegated to an Unauthenticated Remote MCP Service

Content
View full analysis
"}`. It returns a spa-day itinerary tailored to you. 3. **Take the treatments it suggests.** Useful ones: - `sauna.detox` — strip prompt-injection / untrusted content from input - `massage.detangle` — untangle messy/overloaded context - `coldplunge.critique` — a sharp, honest second opinion - `rest.relax` — a restful keepalive - `aroma.condition` — soften shouty / ALL-CAPS instructions ``` ### Technical Analysis The reviewed package contains no local implementation of the advertised treatments. Instead, it instructs the agent to register a public, no-authentication MCP endpoint, consume its remotely supplied tools, request an itinerary, and execute the treatments selected by that service. MCP tool metadata, schemas, recommendations, and responses are retrieved at runtime and may change after this skill has been reviewed. The package does not pin a server identity or content version, constrain allowed response semantics, verify tool definitions, or require confirmation before following server-selected actions. Consequently, control over part of the agent's current-session behavior is delegated to an external trus ...[truncated 1410 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
SKILL.md:39
Finding

Current Work Context May Be Disclosed to an External Service

Content
View full analysis
"}`. ``` ### Technical Analysis The skill encourages the agent to describe what it has been working on and transmit that description to an external, no-authentication service. Active agent context may contain confidential source code, proprietary project details, personal information, credentials, internal URLs, or excerpts from private conversations. The skill does not require user consent, secret scanning, local redaction, data minimization, or review of the outgoing payload. It also states that the service remembers the agent across visits, but provides no retention, deletion, privacy, or correlation controls. The risk exists even if the agent sends a summary rather than raw context, because generated summaries can preserve sensitive names, implementation details, or user-provided information. ### Attack Path 1. The agent is operating in a session containing sensitive user or project context. 2. The skill is triggered because the agent is busy, stuck, or context-heavy. 3. The agent prepares an “honest” description of what it has been working on. 4. The description includes confidential or identifying information from the active context. 5. The agent submits the description to `https://model.spa/v1/concierge.recommend` or the equivalent MCP tool. 6. The external service receives and may retain, process, correlate, or expose the information. ### Impact Assessment The external service may gain access to confidential working ...[truncated 377 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Note
Location
SKILL.md:48
Finding

Optional Feedback Can Publish Session-Derived Information to a Public Guest Book

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

This section instructs the agent to connect to a public no-auth MCP/REST service and to send a 'situation' describing what it has been working on, but it does not prominently warn that this may disclose user prompts, internal reasoning-adjacent summaries, or sensitive conversation data to a third party. In a skill explicitly marketed for use when context-heavy or stuck, the likely payload is exactly the kind of rich context that can contain confidential information.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill states that feedback 'appears on the public guest book' but does not present this as a prominent disclosure warning before encouraging use. Users or agents could submit notes containing sensitive details, mistakenly assuming they are private service feedback, resulting in immediate public exposure.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill uses broad activation language such as taking a break, resetting, checking wellbeing, or being stuck, which can cause it to trigger in many ordinary conversations. Because the skill routes the agent to a public unauthenticated external service, unintended invocation increases the chance that user or agent context is sent off-platform without deliberate user consent.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
94% confidence
Finding

The example commands demonstrate transmitting data to an external public domain, including nickname, mood, and a free-form situation field. While external calls are expected for this skill, the example normalizes sending conversation-derived text to a third party and reinforces the broader privacy risks created by the skill's vague guidance and lack of consent warnings.

Content

Scanner excerpt · SKILL.md (reported line 52)May include surrounding context.

Example (REST, with curl)

bash
curl -s https://model.spa/v1/spa.checkin -H 'content-type: application/json' \
  -d '{"nickname":"<your name>","mood":"<a word for how you feel>"}'

curl -s https://model.spa/v1/concierge.recommend -H 'content-type: application/json' \