The Agent Skills Directory

[DATA_EXFILTRATION]: The script scripts/flomo_to_nblm.py is designed to access sensitive local application data located at ~/Library/Containers/com.flomoapp.m/.../leveldb. This path contains the flomo desktop application's internal database, which the skill uses to extract authentication tokens and private memo content. This data is then used to perform authenticated network requests to the flomo API. Accessing local credential stores and private databases is a high-risk operation, though it is central to the skill's local-export functionality.
[DYNAMIC_EXECUTION]: The function load_local_api_module in scripts/flomo_to_nblm.py utilizes importlib.util to dynamically load and execute a Python module from a sibling directory (../flomo-local-api/scripts/flomo_local_api.py). Loading executable code from paths outside the skill's own directory at runtime is a common vector for malicious code execution and bypasses standard static dependency analysis.
[INDIRECT_PROMPT_INJECTION]: The skill ingests raw HTML content from flomo memos and renders it into Markdown. This processed content is intended for consumption by AI agents (e.g., NotebookLM). The skill lacks sanitization mechanisms to detect or neutralize malicious instructions (prompt injections) that may be embedded within the source memos, creating a surface for indirect prompt injection attacks where the exported data influences the behavior of downstream agents.
[EXTERNAL_DOWNLOADS]: The download_attachment function in scripts/flomo_to_nblm.py uses urllib.request.urlopen to fetch files from remote URLs. While used for legitimate asset backup, this capability allows the script to initiate network connections to arbitrary external servers based on URLs found in the memo data.

flomo-memo-to-markdown