The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is instructed to read local configuration files, specifically user-config.yaml and files in {project-root}/_bmad/, to resolve variables and project knowledge. These files often contain environment-specific metadata or sensitive project structures.
[EXTERNAL_DOWNLOADS]: The instructions require the agent to fetch remote documentation from URLs provided in the bmad-help.csv file (specifically rows marked with _meta). This involves performing network requests to retrieve external content at runtime.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it incorporates untrusted data into the agent context.
Ingestion points: Data is ingested from the local bmad-help.csv file and remote URLs (e.g., llms.txt).
Boundary markers: Absent. The skill does not define delimiters or specific instructions to treat fetched content as untrusted data.
Capability inventory: The skill has the capability to read local files (CSVs, YAMLs, artifacts) and fetch remote data via URLs.
Sanitization: Absent. There are no instructions to sanitize, filter, or validate the content retrieved from remote documentation sources before using it to answer user questions.

bmad-help