Skills
Navigate
skills/akin-ozer/cc-devops-skills/github-actions-validator/Gen Agent Trust Hub

github-actions-validator

Fail

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The scripts/install_tools.sh file downloads and directly runs external shell scripts via bash piping (curl ... | bash) and process substitution (bash <(curl ...)).
  • [EXTERNAL_DOWNLOADS]: Scripts are downloaded at runtime from external repositories (https://raw.githubusercontent.com/nektos/act/master/install.sh and https://raw.githubusercontent.com/rhysd/actionlint/main/scripts/download-actionlint.bash) to set up local binary prerequisites.
  • [DYNAMIC_EXECUTION]: The scripts/validate_workflow.sh file utilizes eval to run dynamically constructed command strings when invoking the local act tool workflow listing and dry-run procedures.
  • [INDIRECT_PROMPT_INJECTION]: The skill contains an input ingestion surface for untrusted external data.
  • Ingestion points: scripts/validate_workflow.sh takes user-specified paths or directories as command-line arguments.
  • Boundary markers: Absent; there are no delimiters or warnings to prevent malicious content within workflow files from influencing tool output.
  • Capability inventory: Spawns subprocesses to invoke actionlint and act, with act potentially interacting with the local Docker daemon.
  • Sanitization: Absent; the script verifies the file path existence but does not sanitize the structural contents of the workflow configurations before processing.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/nektos/act/master/install.sh, https://raw.githubusercontent.com/rhysd/actionlint/main/scripts/download-actionlint.bash - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 15, 2026, 02:25 PM