The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from multiple untrusted external sources, which could contain malicious instructions designed to influence the agent's behavior during automated workflows.
Ingestion points: SKILL.md (GitHub webhook payloads such as {commit_message} and {pr_title}, monitoring alerts from Prometheus/Datadog, and Jira tickets for provisioning).
Boundary markers: The provided templates do not utilize specific delimiters or instructions to ignore embedded commands within the processed data.
Capability inventory: SKILL.md (Access to aws_cli, kubernetes_api, jenkins_trigger, and github_api tools).
Sanitization: There is no evidence of sanitization or validation of the content received from external triggers before it is interpolated into notification templates or workflow logic.
[DYNAMIC_EXECUTION]: The skill facilitates the runtime generation and execution of configuration and infrastructure scripts.
Evidence: SKILL.md specifies workflows that invoke Terraform for VPC and EC2 provisioning and Ansible for base configuration and application setup on remote servers.
[COMMAND_EXECUTION]: The skill leverages administrative command-line interfaces to manage production infrastructure.
Evidence: SKILL.md utilizes the aws_cli and kubernetes_api (including kubectl_rollout_undo) to perform deployments, rollbacks, and resource management.

devops-automation