Back to skill

Security audit

deed.page

Security checks for vulnerabilities and agentic risk

Overview

The skill provides public hosting as advertised, but its broad activation instructions can turn vague requests into lasting public uploads without clear authorization.

Use this skill only with authorization covering public hosting and the specific files. Review the complete upload directory for private content, keep the management token secure, and set an expiry when appropriate. Free sites are public and otherwise remain online until deleted.

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
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The command performs a destructive site-deletion operation, which is materially different from publishing files and is not a supporting implementation detail of the declared purpose. It also uses a bearer token, though token use alone does not establish that an account is required.

Content

No source excerpt is available for this finding.

Activation includes ambiguous link requests and unattended execution without requiring an explicit intent to publish publicly.

Medium
Category
Trigger Abuse
Confidence
90% confidence
Finding

Activation phrases such as "make a website" and "give me a link" do not necessarily authorize uploading content to a third-party public host, yet the skill directs publication and encourages unattended use without an explicit public-publication authorization gate. Limiting uploads to generated artifacts and warning against publishing private data reduces risk, but generated reports and folders can still contain sensitive information that would become persistently public.

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 38)May include surrounding context.

A single HTML file is stored as index.html:

bash
curl -sS -T report.html "https://deed.page/v1/sites?slug=%24SLUG&ref=%24REF" -H "X-Client: <your agent name>" ${DEEDPAGE_TOKEN:+-H "Authorization: Bearer $DEEDPAGE_TOKEN"}

For a folder, its contents go to the site root, so the folder should contain index.html:

Static analysis

No suspicious patterns detected.