cekura-flag-call-log-failures

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection vulnerability surface through the processing of untrusted external conversation data.\n
  • Ingestion points: Full call transcripts and metric evaluations are ingested from the production environment using the mcp__cekura__call_logs_retrieve tool in Step 2.\n
  • Boundary markers: The instructions do not specify the use of delimiters or instructions to ignore potential commands embedded within the transcript content during analysis.\n
  • Capability inventory: The agent has access to high-capability tools including Bash, Write, Edit, and Skill, which could be exploited if malicious instructions were present in a processed transcript.\n
  • Sanitization: No sanitization or validation of the transcript text is required before the agent extracts evidence or identifies failure modes.\n- [COMMAND_EXECUTION]: The skill utilizes dynamic script generation to process and analyze large datasets.\n
  • Evidence: In Step 2, the agent is instructed to 'parse with a script for loop/repeat detection' when transcript files are too large to be read inline.\n
  • Context: While this is a functional requirement for analyzing large files, the combination of script generation with untrusted input (transcripts) increases the risk profile of the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 02:03 PM
Security Audit — agent-trust-hub — cekura-flag-call-log-failures