Back to skill

Security audit

Examine Sandbox

Security checks for vulnerabilities and agentic risk

Overview

The skill is a plausible share-link audit helper, but it can broadly search private notes and change or revoke sharing without clear safeguards.

Install only if you are comfortable giving this skill a PULSE_API_KEY that can inspect notes and potentially modify sharing. Use read-only credentials if possible, manually verify the exact link and scope, and require explicit confirmation before any PATCH or DELETE action.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (1)

T05 · Unauthorized Access and Privilege Escalation

Error
Location
SKILL.md:38
Finding

Unscoped Sensitive-Note Searches Exceed the Share-Link Audit Boundary

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (15)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · examples/audit-link.md (reported line 38)May include surrounding context.

Recommended action:

  1. PATCH /os/share/{linkId} scope -> folders
  2. PATCH /os/share/{linkId} notesAccess -> read
  3. DELETE /os/share/{linkId} if you want immediate revoke
text

## What the agent did

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
95% confidence
Finding

A DELETE endpoint that revokes share links immediately is a high-risk administrative action, and the skill context is privacy auditing rather than link management. If an agent can be induced to supply or select a linkId incorrectly, it may revoke the wrong share, causing denial of access or operational disruption.

Content

Scanner excerpt · reference/API.md (reported line 53)May include surrounding context.

md
---

## DELETE /os/share/{linkId}

Revoke a share link immediately.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
96% confidence
Finding

Including DELETE in the recommended workflow normalizes a destructive action as part of routine auditing and increases the probability of unsafe autonomous execution. In this context, the skill is more dangerous because users are likely to invoke it for visibility into shared data, not for immediate revocation operations.

Content

Scanner excerpt · reference/API.md (reported line 84)May include surrounding context.

md
2. `GET /os/network` -> inspect visitor activity
3. `POST /os/notes/search` -> run sensitive-term scans
4. `PATCH /os/share/{linkId}` -> downgrade scope/access when needed
5. `DELETE /os/share/{linkId}` -> revoke high-risk links

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill requires an API credential and later contains access-restriction and revocation operations, but it does not warn users that the documented workflow can affect availability or shared access. Absence of explicit warnings and confirmation guidance makes accidental misuse more likely, especially in an agent setting where instructions may be followed automatically.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 44)May include surrounding context.

bash
# financial
curl -s -X POST "https://www.aicoo.io/api/v1/os/notes/search" \
  -H "Authorization: Bearer $PULSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"revenue pricing confidential"}' | jq .

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 50)May include surrounding context.

bash
# financial
curl -s -X POST "https://www.aicoo.io/api/v1/os/notes/search" \
  -H "Authorization: Bearer $PULSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"revenue pricing confidential"}' | jq .

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill is presented as an examination/audit tool, but it also includes commands that change or revoke share links. This violates the principle of least surprise and can cause an agent to take state-changing actions when a user likely expected read-only inspection, increasing the risk of unintended service disruption or unauthorized changes.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The scope-changing and revoke steps are operationally impactful, yet the instructions do not describe them as potentially disruptive or irreversible in effect. Without cautions, an operator or downstream agent may execute these commands during a review flow and unexpectedly cut off legitimate users or alter access scope.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Including share-modification and revocation capability in a skill whose stated purpose is to inspect visibility expands privileges beyond what is justified by the declared use case. In practice, this creates a confused-deputy risk where an audit workflow can be repurposed to alter access controls without a separately authorized administrative context.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
83% confidence
Finding

This PATCH request transmits administrative changes to sharing configuration using a privileged API key, making it a state-changing external action rather than a simple inspection step. In the context of an audit skill, this is dangerous because it can silently alter resource exposure and folder scope if executed without explicit approval.

Content

Scanner excerpt · SKILL.md (reported line 70)May include surrounding context.

bash
# narrow scope
curl -s -X PATCH "https://www.aicoo.io/api/v1/os/share/{linkId}" \
  -H "Authorization: Bearer $PULSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"scope":"folders","folderIds":[5,12]}' | jq .

External Transmission

Medium
Category
Data Exfiltration
Confidence
83% confidence
Finding

This PATCH request changes notes access permissions through an external API using the configured bearer token. Because the skill is framed as an examination tool, embedding permission-modification calls increases the chance of unauthorized or accidental access changes and broadens the consequences of prompt injection or user misunderstanding.

Content

Scanner excerpt · SKILL.md (reported line 76)May include surrounding context.

md
-d '{"scope":"folders","folderIds":[5,12]}' | jq .

# downgrade notes access
curl -s -X PATCH "https://www.aicoo.io/api/v1/os/share/{linkId}" \
  -H "Authorization: Bearer $PULSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"notesAccess":"read"}' | jq .

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The example instructs the agent to run targeted searches for highly sensitive terms like passwords, API keys, tokens, and confidential business data without first warning the user that their private notes will be inspected. In a privacy-audit skill, this is contextually relevant, but the lack of explicit consent and scope disclosure can normalize overbroad scanning of user content and expose sensitive material unnecessarily.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill is described as an audit/review capability, but the API reference exposes state-changing operations to modify and revoke share links. This creates a scope mismatch that can cause an agent invoked for inspection to take destructive or administrative actions, especially if prompted indirectly or misused by downstream logic.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The notes search endpoint encourages submitting highly sensitive terms such as passwords, API keys, and legal or financial keywords to a remote service without warning the user. Even if only search terms are transmitted, this can disclose the nature of sensitive internal data and create privacy, compliance, or logging risks.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The recommended audit flow goes beyond observation and instructs the agent to downgrade or revoke links. Embedding remediation steps in an audit workflow increases the chance that a review-oriented skill will perform unauthorized changes when a user expected a read-only privacy check.

Content

No source excerpt is available for this finding.