The Agent Skills Directory

[DATA_EXPOSURE]: The script scripts/task-add-run.sh contains a load_dotenv function that explicitly reads and parses the .env file located in the project root.
Evidence: The script calls load_dotenv "$ROOT_DIR/.env" to populate environment variables such as PFLOW_TASKS_MDTODO_FILE.
Risk: Accessing .env files is flagged because they frequently contain sensitive credentials, although in this context it is used for local configuration.
[INDIRECT_PROMPT_INJECTION]: The skill takes user-supplied titles and descriptions and writes them directly into a markdown file (todo.md) without sanitization or boundary markers.
Ingestion points: User input is passed through the --title and --description flags to the execution script.
Capability: The skill has file-write permissions to update the task list.
Sanitization: There is no validation or escaping of the user content before it is appended to the markdown file, which could allow malicious instructions to be stored and later interpreted by an agent reading the tasks.
[COMMAND_EXECUTION]: The script executes the mdtodo command-line interface to add new tasks.
Evidence: The script uses a bash array cmd=(mdtodo add ...) and executes it via "${cmd[@]}".
Risk: While the execution method is secure against argument injection, the skill relies on the presence and behavior of an external CLI tool.

pflow-task-add