NYC

sentry-create-alert

Warn

Audited by Gen Agent Trust Hub on Feb 18, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • COMMAND_EXECUTION (MEDIUM): The skill constructs multiple shell commands using variables such as {org}, {token}, and {payload}. In Phase 4, the execution of curl -d '{payload}' is particularly risky if the agent does not properly escape the payload string, potentially allowing for command injection if an attacker can influence the configuration values.\n- CREDENTIALS_UNSAFE (MEDIUM): The skill explicitly requests and utilizes a Sentry Organization Auth Token. Handling this token within shell commands and environment variables increases the risk of credential exposure in system logs, shell history, or process monitoring tools.\n- INDIRECT_PROMPT_INJECTION (LOW): (Category 8) The skill ingests untrusted data from Sentry API endpoints in Phase 2 (member emails, team names, integration names). \n
  • Ingestion points: SKILL.md (Phase 2 API calls to /members/, /teams/, and /integrations/).\n
  • Boundary markers: Absent; the agent is not instructed to ignore instructions embedded in the API responses.\n
  • Capability inventory: Subprocess execution (curl, python3), network write operations (POST, DELETE).\n
  • Sanitization: Absent; the skill does not include logic to sanitize or validate the strings returned from the Sentry API before they are processed by the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 18, 2026, 02:07 AM