The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The script scripts/flomo_local_api.py contains a hardcoded authentication secret (API_SECRET = "dbbc3dd73364b4084c3a69346e0ce2b2"). Additionally, it implements a mechanism to harvest an access_token by reading raw data files from the Flomo desktop app's internal storage at ~/Library/Containers/com.flomoapp.m/Data/Library/Application Support/flomo/Local Storage/leveldb.
[COMMAND_EXECUTION]: The skill relies on shell command execution to run Python scripts that perform network operations, file system reads, and file system writes.
[DATA_EXFILTRATION]: The skill accesses personal memo data from the Flomo API and can exfiltrate this information to the local filesystem by writing it to the ~/download/ directory during the export-monthly command.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests untrusted data from the Flomo API.
Ingestion points: Memo content and tag data are fetched from flomoapp.com in scripts/flomo_local_api.py via api_get calls.
Boundary markers: The skill lacks explicit delimiters or instructions to the agent to treat fetched memo content as untrusted data.
Capability inventory: The skill possesses network read/write capabilities (api_get/api_put) and local file write capabilities (Path.write_text).
Sanitization: No sanitization or validation is performed on the fetched content to prevent embedded instructions from influencing agent behavior.

flomo-local-api