Skills
skills/allium-labs/skills/dune-to-allium/Gen Agent Trust Hub

dune-to-allium

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local Python scripts allium_query.py, dune_query.py, and compare_results.py via uv run to perform data retrieval and validation tasks.\n- [EXTERNAL_DOWNLOADS]: The skill communicates with the official API endpoints api.allium.so and api.dune.com. These are recognized as well-known service providers necessary for the skill's primary functionality.\n- [DATA_EXFILTRATION]: The scripts read Allium and Dune API keys from ~/.allium/credentials and .env files. These credentials are used appropriately to authenticate requests to the respective official services.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface (Category 8) due to its handling of untrusted SQL queries which are subsequently executed.\n
  • Ingestion points: User-provided SQL text and query IDs processed in the workflow defined in SKILL.md.\n
  • Boundary markers: Absent; the instructions do not include specific delimiters or warnings to ignore instructions embedded within the SQL data.\n
  • Capability inventory: The skill can execute arbitrary SQL on the Allium Explorer API through the allium_query.py script.\n
  • Sanitization: Absent; while the skill performs structural SQL transformations, it does not explicitly sanitize the input for potential instruction injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 08:40 AM