truefoundry-observability
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides tools and instructions to download and view application logs, which are external, untrusted data sources. An attacker who can control log output could potentially embed malicious instructions designed to influence the agent's behavior during log summarization or analysis.
- Ingestion points: Application logs are ingested via the
tfy_logs_downloadtool call and thetfy-api.shscript. - Boundary markers: The skill includes a warning for the agent to review logs for sensitive content (secrets, PII) before responding, but it lacks explicit instructions to treat log content as untrusted data or use delimiters to prevent prompt injection.
- Capability inventory: The skill has capabilities to modify source code (instrumentation), install packages (
pip,npm), and execute shell scripts (tfy-api.sh). - Sanitization: No specific sanitization or filtering for embedded instructions is mentioned for the log processing workflow.
Audit Metadata