budgeting-planning
Warn
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The SKILL.md file contains a 'Companion check' section that instructs the AI agent to execute directory listing commands (
ls) on internal system paths including~/.claude/skills/,~/.agent/skills/, and~/.agents/skills/.\n- [DATA_EXFILTRATION]: The instruction to list internal directories serves as environment reconnaissance, allowing the agent to identify and expose the presence of other installed tools, custom scripts, or private extensions to its operational context.\n- [EXTERNAL_DOWNLOADS]: The skill documentation encourages the acquisition of additional extensions from the 'AbsolutelySkilled' GitHub repository, which is not identified as a trusted vendor or well-known service.\n- [REMOTE_CODE_EXECUTION]: The recommended installation method (npx skills add) involves downloading and executing code from a remote third-party source, which could facilitate the execution of arbitrary scripts from an unverified repository.\n- [PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection because it is designed to ingest and process untrusted external data (financial plans and budgets) while possessing the capability to execute shell commands, yet it lacks explicit boundary markers or input sanitization protocols.\n - Ingestion points: Financial data processed during budget building and variance analysis (SKILL.md).\n
- Boundary markers: None identified in the skill instructions.\n
- Capability inventory: Shell command execution via tool use (
ls,npx).\n - Sanitization: No validation or sanitization of input data is defined before processing.
Audit Metadata