The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The script scripts/flomo_to_nblm.py accesses the local flomo desktop application's storage at ~/Library/Containers/com.flomoapp.m/.../leveldb to extract authentication tokens.
[REMOTE_CODE_EXECUTION]: The script scripts/flomo_to_nblm.py contains a dynamic loading mechanism in the load_local_api_module function. It uses importlib.util.spec_from_file_location and spec.loader.exec_module to load and execute a Python script located at ../../flomo-local-api/scripts/flomo_local_api.py. This allows for execution of code residing outside the skill's own file tree.
[EXTERNAL_DOWNLOADS]: The download_attachment function in scripts/flomo_to_nblm.py uses urllib.request.urlopen to fetch data from remote URLs, which are then saved to the local file system using pathlib.Path.write_bytes.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It ingests untrusted memo content via the local API and converts it to Markdown for use by an agent or NotebookLM. 1. Ingestion points: scripts/flomo_to_nblm.py fetches content via local_api.fetch_all_memos(). 2. Boundary markers: None identified; the script does not use delimiters or instructions to ignore embedded commands in the memo content. 3. Capability inventory: The skill has file-writing capabilities, network access, and dynamic execution capabilities. 4. Sanitization: It uses BeautifulSoup for HTML parsing, but lacks logic to sanitize or escape instructions that might be embedded in the memo text.

flomo-memo-to-markdown