The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests content from external [會議記錄] Google Docs to update project management Google Sheets. Maliciously crafted content within the meeting minutes could potentially influence the agent's behavior or corrupt the target spreadsheets.\n
Ingestion points: Uses get_doc_content and get_doc_as_markdown in Step 2 to extract text and tables from Google Docs.\n
Boundary markers: There are no explicit instructions to delimit the untrusted content or warn the agent to ignore embedded instructions within the parsed text.\n
Capability inventory: The skill possesses the ability to execute shell commands (python3 ./scripts/pm-schema.py), modify Google Sheets (mcp__google-workspace__modify_sheet_values), and edit Docs (batch_update_doc).\n
Sanitization: While value_input_option="RAW" is used for Sheets to prevent formula injection, the text content itself is not sanitized against indirect prompt injection attacks.\n- [DYNAMIC_EXECUTION]: The scripts/pm-schema.py file uses the Python ast module to perform static analysis and runtime evaluation of constants in other project files. This involves interpreting Python code structures like list comprehensions and formatted strings at runtime.\n- [EXTERNAL_DOWNLOADS]: The documentation includes an installation command npx skills add https://github.com/peter-tu-zynkr/zynkr-skill-builder which references the author's public repository.

project-minutes-sync