Back to skill

Security audit

DeepRead Insurance Documents

Security checks for vulnerabilities and agentic risk

Overview

The skill is a disclosed cloud document-processing integration, but it needs Review because it uploads complete sensitive insurance and EOB documents to a third-party service without enough privacy, authorization, or compliance guidance.

Install only if you are authorized to send the insurance or EOB documents to DeepRead and its processing providers. Treat uploaded files as unredacted disclosures to a third-party service, including when using the redaction feature. Review DeepRead's privacy, retention, deletion, residency, subprocessors, and compliance terms before using it with regulated or confidential records, and minimize or redact locally where possible.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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 (2)

other

Error
Location
SKILL.md:133
Finding

Unredacted Insurance Documents Are Disclosed to a Third-Party Processing Service

Content
View full analysis
This skill instructs the agent to POST documents to `https://api.deepread.tech` and poll for results. No system files are modified. ``` ```python with open("claim.pdf", "rb") as f: job = requests.post( f"{BASE}/v1/process", headers=headers, files={"file": f}, data={"schema": schema}, ).json() ``` ```python # Step 1: Extract the claim data you need with open("claim.pdf", "rb") as f: extract_job = requests.post( f"{BASE}/v1/process", headers=headers, files={"file": f}, data={"schema": schema}, ).json() # Step 2: Redact PII from the original with open("claim.pdf", "rb") as f: redact_job = requests.post( f"{BASE}/v1/pii/redact", headers=headers, files={"file": f}, ).json() ``` ### Technical Analysis The Skill directs the agent to upload complete insurance documents to `https://api.deepread.tech`. Claims, policies, Explanations of Benefits, and loss reports can contain highly sensitive information, including claimant names, addresses, medical information, policy and claim identifiers, financial amounts, property details, and incident descriptions. Although remote document submission is relevant to the declared cloud-processing functionality, the workflow does not require explicit, informed approval immediately before uploading a particular document. It also does not provide instructions for verifying that the user is authorized to disclose the document or explain data retention, deletion, residency, subprocessors, or regulatory controls. The redaction feature does not protect information from disclosure to DeepRead because redaction is performed server-side. The unredacted source document must first be transmitted to the external service. In the co ...[truncated 1905 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:206
Finding

Server-Controlled Download URL Is Fetched Without Scheme or Destination Validation

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill directs users to upload insurance claims, policies, EOBs, and loss reports—documents that commonly contain highly sensitive PII, financial, and health-related data—to a third-party API without a prominent privacy/security warning or guidance on data handling. In this context, omission of an explicit warning can lead users to transmit regulated or confidential data without understanding retention, sharing, jurisdiction, or compliance implications, increasing the likelihood of privacy violations.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

The skill explicitly instructs the agent/user to transmit insurance PDFs to an external endpoint at api.deepread.tech. Because the documents described include claimant identity data, claim details, financial information, and potentially medical EOB content, this external transmission meaningfully expands the data exposure boundary and can create confidentiality, compliance, and third-party risk if users are unaware or controls are insufficient.

Content

Scanner excerpt · SKILL.md (reported line 168)May include surrounding context.

cURL

bash
curl -s -X POST https://api.deepread.tech/v1/process \
  -H "X-API-Key: $DEEPREAD_API_KEY" \
  -F "file=@claim.pdf" \
  -F 'schema={"type":"object","properties":{"claim_number":{"type":"string","description":"Claim number"},"policy_number":{"type":"string","description":"Policy number"},"claimant_name":{"type":"string","description":"Claimant name"},"date_of_loss":{"type":"string","description":"Date of loss"},"loss_type":{"type":"string","description":"Type of loss"},"estimated_damages":{"type":"number","description":"Total damages"},"status":{"type":"string","description":"Claim status"}}}'