The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to find and parse configuration files that commonly store sensitive credentials or environment variables, specifically targeting files like terraform.tfvars, local.settings.json, and general codebase search for API integrations. While this access is intended to identify infrastructure sizes and providers for cost modeling, it exposes potentially sensitive data to the AI's processing context.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and interprets untrusted content from the project's source code and configuration files, creating an attack surface for indirect prompt injection.
Ingestion points: Reads various project files including firebase.json, serverless.yml, *.tf, *.tfvars, and others across the directory tree.
Boundary markers: There are no specific instructions or delimiters provided to the agent to help it distinguish between legitimate data and potential malicious instructions embedded in the codebase. Furthermore, the instruction to 'not ask the user questions' increases the risk by encouraging autonomous execution without oversight.
Capability inventory: The agent has the ability to read files, write reports to the filesystem (docs/cost-analysis.md), and potentially access the network for pricing information.
Sanitization: The skill does not perform any sanitization or validation of the ingested file content before it is processed.

cost-analysis