The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions are focused on functional parsing and data mapping. There are no attempts to override safety filters, bypass constraints, or extract system prompts.
[DATA_EXFILTRATION]: Data operations are restricted to searching, creating, and updating rows within the user's own Notion database using the official Notion connector. No external network requests to third-party domains or sensitive file access patterns were found.
[COMMAND_EXECUTION]: The skill does not use shell commands, subprocesses, or platform-level command execution (like the ! syntax).
[REMOTE_CODE_EXECUTION]: No remote scripts are downloaded or executed. The skill relies entirely on structured tool calls provided by the Notion integration.
[CREDENTIALS_UNSAFE]: No hardcoded API keys, tokens, or sensitive environment variable paths are present. The skill correctly instructs the user to utilize the official connector's authentication.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (activity reports). While it lacks explicit boundary markers, the risk is mitigated by its design: it maps input into a rigid, structured database schema (Activity, Minutes, Date) and uses a 'To-confirm' status for ambiguous entries, preventing instructions inside user text from being executed as commands.

time-ledger