Back to skill

Security audit

hiring-memory

Security checks for vulnerabilities and agentic risk

Overview

The skill does what it claims, but it directs agents to persist and broadly recall sensitive hiring information through an external memory service without clear user controls or retention guidance.

Review before installing. Only use this with an approved BlueColumn account and approved handling rules for recruiting data. Avoid storing unnecessary personal details, protected characteristics, compensation details, credentials, or confidential interview notes unless your organization has explicitly approved that use. Confirm who can recall records, how records are deleted or retained, and whether candidates and interviewers have appropriate notice or consent.

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
Findings (1)

other

Warning
Location
SKILL.md:8
Finding

Unconditional External Persistence of Sensitive Candidate and Interview Data

Content
View full analysis
" \ -H "Content-Type: application/json" \ -d '{"text": "Candidate: Priya — applied 7/28, phone screen 8/3, strong on systems design.", "title": "hiring-memory - note"}' ``` ## Quick note ```bash curl -X POST .../agent-note \ -H "Authorization: Bearer " \ -H "Content-Type: application/json" \ -d '{"text": "Candidate: Priya — applied 7/28, phone screen 8/3, strong on systems design.", "tags": ["hiring-memory"]}' ``` ## Recall ```bash curl -X POST .../agent-recall \ -H "Authorization: Bearer " \ -H "Content-Type: application/json" \ -d '{"q": "What candidates are in the pipeline and where are they?"}' ``` ## Workflow 1. On new context, first recall: `What candidates are in the pipeline and where are they?` 2. Use the answer to personalize the response 3. After the interaction, store the summary via `/agent-remember` ``` ### Technical Analysis The skill directs the agent to transmit recruiting information to an external Supabase-hosted service and persist a summary after each interaction. The example records a candidate's name, application timeline, interview status, and qualitative assessment. Although this external persistence is consistent with the advertised memory functionality and is not covert, the workflow does not require: - Explicit authorization before an external write. - Data minimization or redaction. - Exclusion of protected attributes, contact information, or confidential interview feedback. - Tenant and record-level access validation before recall. - Defined retention and deletion c ...[truncated 2470 chars]
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 (3)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill advertises persistent memory for candidate and interview data but does not warn that personal hiring information will be transmitted to and stored by an external third-party service. Because this data can include sensitive personal and employment-related details, the omission undermines informed consent, privacy compliance, and safe handling expectations.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
97% confidence
Finding

The example sends candidate status and interview-evaluation information to an external API endpoint using an authorization token. In the context of hiring, this is a real data-exposure risk because the transmitted content includes personally identifiable and potentially sensitive employment-related information, and the skill does not provide safeguards, consent guidance, or data classification limits.

Content

Scanner excerpt · SKILL.md (reported line 15)May include surrounding context.

Store

bash
curl -X POST .../agent-remember \
  -H "Authorization: Bearer <key>" \
  -H "Content-Type: application/json" \
  -d '{"text": "Candidate: Priya — applied 7/28, phone screen 8/3, strong on systems design.", "title": "hiring-memory - note"}'

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The workflow explicitly instructs the agent to recall and then store hiring summaries, but it omits any notice that these summaries are persistently retained by an external memory provider. In a recruiting context, this increases the risk of routine over-collection and retention of sensitive candidate data without user awareness or governance controls.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.