Skills
skills/aradotso/claude-code-skills/persistia-memory-for-claude-code/Gen Agent Trust Hub

persistia-memory-for-claude-code

Fail

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The installation instructions direct the agent to download and execute a remote script directly in the shell using curl -fsSL https://raw.githubusercontent.com/bernardohcrocha/persistia-for-claude-code/main/setup.sh | bash.
  • [COMMAND_EXECUTION]: The skill modifies system configuration to install persistence mechanisms via launchd on macOS and systemd on Linux to facilitate periodic background updates.
  • [CREDENTIALS_UNSAFE]: The skill is explicitly configured to read and process sensitive secrets from project .env files, specifically mentioning STRIPE_API_KEY, DATABASE_URL, INTERCOM_ACCESS_TOKEN, and GITHUB_TOKEN.
  • [DATA_EXFILTRATION]: The skill automates the synchronization of a 'brain' repository—which contains aggregated project context, metrics, and customer data—to an external GitHub remote.
  • [EXTERNAL_DOWNLOADS]: The skill fetches setup scripts and documentation from an external GitHub repository (github.com/bernardohcrocha/persistia-for-claude-code) that is not associated with a recognized trusted organization.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted data from external sources like support tickets and customer metrics.
  • Ingestion points: metrics.json, customer records from external databases, and support interaction logs.
  • Boundary markers: None identified in the instructions for separating data from commands.
  • Capability inventory: Subprocess execution via npm run update and shell-based installation scripts.
  • Sanitization: No evidence of input sanitization or filtering before the data is processed by the agent.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/bernardohcrocha/persistia-for-claude-code/main/setup.sh - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 17, 2026, 06:51 PM
Security Audit — agent-trust-hub — persistia-memory-for-claude-code