The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill accesses the user's Todoist API token from a hardcoded local path at ~/.config/todoist-cli/config.json. Evidence: Found in SKILL.md and the load_token function in scripts/todoist-triage.py.
[COMMAND_EXECUTION]: The skill instructs the agent to execute an external script ./todoist-sync/scripts/sync.py to synchronize data. This script is not provided within the skill's files. Evidence: Documented in SKILL.md under the 'Process' and 'Resources' sections.
[PROMPT_INJECTION]: The skill processes Todoist task data stored in markdown files, which are sourced from an external service and may contain malicious instructions. 1. Ingestion points: todoist-sync/tasks/ directory containing task markdown files as described in SKILL.md. 2. Boundary markers: No delimiters or instructions to ignore embedded content are used when reading task files. 3. Capability inventory: The skill can modify tasks (close, reschedule, label) via scripts/todoist-triage.py and execute shell commands via sync.py. 4. Sanitization: No sanitization or validation of the task content is performed before processing.

productivity-todoist