The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill reads configuration data from local files (.agents.yml and .agents.local.yml). If these files are sourced from an untrusted repository or modified via a malicious pull request, they could contain embedded instructions designed to manipulate the agent's logic when it retrieves these values.
Ingestion points: The scripts/config_reader.sh script reads data directly from .agents.yml and .agents.local.yml.
Boundary markers: Absent. There are no instructions or delimiters used to warn the agent that the configuration values might contain untrusted natural language instructions.
Capability inventory: The skill executes a local shell script that uses yq to parse YAML; the resulting values are then used by the agent to determine its tech stack, workflow, or task management behavior.
Sanitization: No sanitization or validation is performed on the values retrieved from the YAML files before they are returned to the agent's context.

config-reader