optimizing-cloud-costs
Warn
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The file
scripts/generate_cost_report.pycontains agenerate_scriptmethod that writes a string directly to a shell script and applies executable permissions usingchmod(0o755). This functionality allows for the dynamic creation and execution of scripts based on potentially unverified or manipulated inputs. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from cloud billing and resource management APIs, including resource tags and item descriptions, without implementing sanitization or boundary markers.
- Ingestion points: External data enters the context via cloud provider cost and usage reports as described in
SKILL.md(Steps 1 and 3). - Boundary markers: There are no delimiters or explicit instructions to ignore embedded commands within the processed cloud metadata.
- Capability inventory: The skill possesses extensive modification capabilities via
Bash(aws:*),Bash(gcloud:*), andBash(az:*)tools, along with generic file write and edit permissions. - Sanitization: No input validation or escaping logic is present to handle malicious strings embedded in cloud resource names or tags.
Audit Metadata