The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to execute various shell commands using curl, aws, gcloud, and wrangler to configure cloud-provider integrations and telemetry sinks.
[PRIVILEGE_ESCALATION]: The integration process involves the creation of IAM roles with CAPABILITY_NAMED_IAM via CloudFormation stacks, which is a necessary step for provisioning cross-account observability permissions.
[REMOTE_CODE_EXECUTION]: The skill deployment instructions involve fetching and executing remote infrastructure templates and function code from vendor-owned storage buckets (firetiger-public) for AWS CloudFormation and Google Cloud Functions.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local configuration files (e.g., package.json, requirements.txt, .git/config) to identify the project technology stack.
Ingestion points: Step 3 of the setup process directs the agent to scan specific configuration files to identify telemetry sources and framework identifiers.
Boundary markers: No explicit delimiters or safety instructions are provided to the agent for handling the content of these files during the detection process.
Capability inventory: The agent can execute shell commands, interact with MCP tools for identity management, and deploy persistent cloud resources.
Sanitization: The skill does not define methods for sanitizing or validating the contents of the detected configuration files before they are interpreted by the agent.

firetiger-setup