ecosystem-ecologist

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides a detailed persona for an Ecosystem Ecologist, focusing on domain-specific methodologies such as eddy covariance, biogeochemistry, and process modeling. It lists legitimate scientific software and community resources (e.g., REddyProc, AmeriFlux, NEON) without any suspicious execution patterns.
  • [EXTERNAL_DOWNLOADS]: The skill references numerous official scientific data repositories and tool websites, including AmeriFlux (ameriflux.lbl.gov), FLUXNET (fluxnet.org), ICOS, and ORNL DAAC. These are well-known, trusted resources within the scientific community.
  • [REMOTE_CODE_EXECUTION]: No remote code execution patterns, such as piped shell commands (e.g., curl | bash), were found. The skill mentions standard data processing packages (e.g., xarray, amerifluxr, bigleaf) but does not include scripts to install or execute them from untrusted sources.
  • [DATA_EXFILTRATION]: There are no indicators of data exfiltration. The network references are for data acquisition and community standards, and there is no logic to read sensitive local files or credentials.
  • [PROMPT_INJECTION]: The skill uses persona-setting instructions ("You are an experienced ecosystem ecologist...") which are standard for AI agent skills and do not attempt to bypass safety guidelines or override system prompts maliciously.
  • [OBFUSCATION]: A thorough check for Base64, zero-width characters, homoglyphs, and hidden text patterns revealed no obfuscated content.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 02:19 PM