add-vault-note

Warn

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs dynamic code generation by instructing the agent to modify the file eth_defi/vault/flag.py. It requires creating new Python constants and updating the VAULT_FLAGS_AND_NOTES dictionary entries based on user-controlled input (Message and Vault address). Writing user-controlled strings directly into executable source files is a high-risk pattern.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because user-provided strings are directly inserted into Python source code without sanitization.
  • Ingestion points: User-provided 'Message', 'Vault address', and 'Flags' parameters defined in SKILL.md.
  • Boundary markers: No delimiters or instructions to ignore embedded commands are present to prevent user input from breaking out of the string literal context.
  • Capability inventory: The agent has the capability to write to the filesystem (eth_defi/vault/flag.py), execute the ruff formatter, and open pull requests to commit these changes.
  • Sanitization: There are no instructions to escape, validate, or filter the user-provided message before it is interpolated into the Python code.
  • [EXTERNAL_DOWNLOADS]: The skill references a JSON metadata file from https://top-defi-vaults.tradingstrategy.ai/top_vaults_by_chain.json. This is a vendor-owned resource used for legitimate data retrieval.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 10, 2026, 07:49 PM