Security audit
Security checks for vulnerabilities and agentic risk
This package is labeled as a habit tracker, but the artifacts implement and document a generic local data-logging tool that stores, searches, and exports arbitrary user input.
Install only if you are comfortable with a mislabeled local data logger. Do not enter sensitive health, routine, or personal information unless you have reviewed the stored files under ~/.local/share/habit and understand that entries may be searchable and exportable locally.
The skill is presented as a habit-building tool, but the documented behavior is a generic data-processing/logging CLI with dynamic shell-like placeholders and unrelated pipeline operations. This mismatch can mislead users or orchestrating agents into invoking capabilities outside the declared scope, enabling inappropriate data handling, logging, export, or command routing under a trusted productivity-themed label.
The command table documents a generic data-processing interface instead of habit-tracking actions, despite the skill claiming to manage habits, streaks, reminders, and progress. In an agent ecosystem, this is dangerous because tool-selection logic may trust the metadata and expose users to unintended ingestion, querying, exporting, or profiling operations against local data.
The skill is presented as a habit-building tool, but its implemented interface is a generic data ingestion/logging toolkit with broad commands unrelated to habits. This mismatch is dangerous because users and calling agents may trust it with sensitive habit or wellness data under false assumptions, while the tool primarily persists arbitrary input and enables later search/export of that data.
The documentation markets a simple, clean habit-management experience, but the exposed interface includes broad data-tooling commands inconsistent with that promise. This increases the chance that users or agents will overtrust the skill and use it in contexts where unrelated data operations are unsafe or privacy-invasive.
Vague command descriptions like 'Ingest', 'Transform', 'Query', and 'Filter' are underspecified and conceal what data is accepted, how it is processed, and what side effects occur. In a tool-consuming agent environment, this ambiguity can cause unsafe invocation, accidental data exposure, or misuse of broad operations beyond the user's intent.
The inline documentation brands the script as a habit utility while the code behaves as a generic data toolkit. Mislabeling a tool's purpose can cause operators or upstream agents to route sensitive user data into a component with broader persistence and retrieval capabilities than expected, increasing the risk of unintended collection and disclosure.
The export functionality aggregates all stored log data into json, csv, or txt files without clearly advertising that capability in a security-focused way. This increases the chance that sensitive accumulated user data can be easily materialized into portable files and exposed through syncing, accidental sharing, or other local access paths.
User-provided input is written verbatim to persistent local log files and mirrored into a history log without a clear upfront warning or consent flow. In the context of a habit skill, users may submit sensitive health, routine, or personal productivity details, creating a privacy risk if the local machine is shared, backed up, or later inspected by other tools.
No suspicious patterns detected.
