The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses sensitive global configuration paths and executes commands that reveal identity and infrastructure details.
Evidence: Checks for the existence of ~/.aws/config and ~/.config/gcloud/configurations/ to determine if cloud CLIs are configured.
Evidence: Executes gh auth status to retrieve GitHub authentication information.
Evidence: Executes kubectl config get-contexts to list Kubernetes cluster information.
Evidence: Runs git remote -v which reveals repository source URLs.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the repository's source files, which could be manipulated to influence the audit summary or recommendations.
Ingestion points: Reads .tf (Terraform) files via grep and inspects package.json, requirements.txt, and pyproject.toml manifests.
Boundary markers: Absent; the skill does not use delimiters or instructions to ignore embedded content in the ingested data.
Capability inventory: The skill has the ability to read local files, execute shell commands, and query external data via internal SubImage tools (subimageListModules, subimageRunCypher).
Sanitization: Absent; there is no evidence of filtering or escaping the data extracted from the repository files.
[COMMAND_EXECUTION]: The skill executes various shell commands to perform its auditing workflow.
Evidence: Uses grep, ls, kubectl, gh, git, awk, and sort to collect environment and repository signals.
[EXTERNAL_DOWNLOADS]: The skill provides links to external resources for documentation and module configuration.
Evidence: References and links to documentation on SubImage's official portal (app.subimage.io).

improve-subimage-coverage