Security audit
Security checks for vulnerabilities and agentic risk
This skill is a local note summarizer that reads a disclosed research notes file and prints summaries/search results, with no evidence of hidden persistence, network exfiltration, or destructive behavior.
Install only if you are comfortable with terminal output showing contents, tags, previews, and derived keywords from your local research_db.json. Avoid running it in shared logs or terminals if those notes contain sensitive information.
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
note_processor.py extract experiment conclusion
## Key Point Detection
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
# Add task to summarize research
task_runner.py add "Summarize experiment results" "documentation"
# When complete
The skill documentation states it works with research_db.json and demonstrates reading stored research notes, but it does not warn users that summaries, keyword extraction, and search may expose sensitive note contents in terminal output. For a note-analysis skill handling accumulated research data, a brief disclosure about potentially sensitive content visibility would be expected.
The limitations section states the stop word list is "English-focused," which implies the skill behavior is tuned to a specific language. The documentation does not present this as an explicit user choice or clearly justified locale-specific constraint, which can conflict with language/locale neutrality expectations.
The module usage text on L006 advertises extract <topic> <keyword>, which implies extracting data from notes, but the corresponding implementation in search_topic only searches for occurrences of a keyword and prints matching note snippets. This is an intent/documentation mismatch rather than a security-relevant capability expansion.
This code accesses a local research database in the user's home directory and later prints note previews, summaries, keywords, and matched content to the terminal. Although reading and displaying notes is part of the tool's purpose, there is no explicit disclosure that potentially sensitive research content from the local workspace will be surfaced in command output.
The help output labels the extract command as Search within topic, while the command name and module usage text frame it as extraction. This inconsistency creates conflicting intent signals in the skill's own documentation about what the command actually does.
No suspicious patterns detected.