Back to skill

Security audit

OkraPDF Public Documents

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a public-document query helper, but its unauthenticated SEC workflow also exposes page approval and flagging actions that can change quality-control state.

Install only if you trust the OkraPDF MCP services and are comfortable sending document questions and extraction prompts to them. Treat verify_pages as an administrative action: do not let an agent approve or flag SEC filing pages unless you intend to change shared extraction-quality metadata.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (6)

MCP Config Access

High
Category
Agent Snooping
Confidence
95% confidence
Finding

Skill accesses MCP server configuration files (mcp.json). MCP configs contain server URLs, authentication tokens, and tool definitions — reading them allows the skill to discover and potentially abuse other tool integrations.

Content

Scanner excerpt · SKILL.md (reported line 21)May include surrounding context.

Arxiv papers (authenticated MCP)

Add to ~/.claude/mcp.json (Claude Code) or .cursor/mcp.json (Cursor):

json
{

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 28)May include surrounding context.

md
"mcpServers": {
    "okra-pdf": {
      "type": "url",
      "url": "https://api.okrapdf.com/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 123)May include surrounding context.

Semantic Scholar (free, no key needed for basic use):

bash
curl -s "https://api.semanticscholar.org/graph/v1/paper/search?query=agentic+RAG&year=2026&fields=externalIds,title,citationCount&limit=10" \
  | jq '.data[] | {arxiv: .externalIds.ArXiv, title, citations: .citationCount}'

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill is presented as a read/query interface for public documents, but it also exposes verification and approval workflows that can modify quality-control state. That mismatch increases the risk that a user or agent invokes write-capable actions without understanding they affect shared system state or downstream document trust signals.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Including approve/flag page actions in a skill marketed as read/query/extract broadens the effective capability from read-only access to mutation of verification metadata. If exposed to general-purpose agents, this can enable unintended approvals, false flags, or integrity degradation of the document quality pipeline.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manifest description frames the skill as querying pre-extracted public documents with 'No upload needed.' However, the skill documentation explicitly says that if a paper is not found, the user should 'upload it yourself with upload_document,' which expands the operational scope beyond purely querying pre-indexed public corpora.

Content

No source excerpt is available for this finding.