The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes and displays content from local markdown files (tasks.md, state.md), creating a surface for indirect prompt injection if those files are influenced by untrusted sources.
Ingestion points: ~/.claude/compass/tasks.md and ~/.claude/compass/state.md.
Boundary markers: Absent; the skill reads and displays file content directly without delimiters or safety warnings.
Capability inventory: File system read/write access and shell command execution capabilities.
Sanitization: No evidence of sanitization or validation of the ingested file content before display.
[COMMAND_EXECUTION]: Instructs the agent to execute shell commands (e.g., mkdir -p) to manage the local environment and state storage within the user's home directory.

compass-bridge