The Agent Skills Directory

[Prompt Injection] (SAFE): No malicious override or bypass patterns were detected. The skill uses standard instructional language to define its task-capturing behavior.
[Data Exposure & Exfiltration] (SAFE): The skill only interacts with a local tasks.md file. No network requests (curl, wget, fetch) or access to sensitive directories (like ~/.ssh or .env) were found.
[Remote Code Execution] (SAFE): There are no patterns involving the download or execution of remote scripts or the installation of third-party packages.
[Indirect Prompt Injection] (LOW): The skill ingests untrusted user input in the form of task descriptions and appends them directly to a file. This represents a potential injection surface if another tool subsequently reads and executes instructions from tasks.md without sanitization.
Ingestion points: User-provided [task description] extracted during the Parse Input step in SKILL.md.
Boundary markers: None; the task text is appended as a markdown list item.
Capability inventory: File-write (append) to tasks.md.
Sanitization: No escaping or validation is performed on the task description before it is written to the file.

task