The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash open command to automatically launch the generated HTML file in the user's default browser. This is an expected behavior for this skill's functionality, but it is the final step in the risk chain involving untrusted content.
[DATA_EXPOSURE]: The skill explicitly gathers information from the session, including conversation history, code changes, and execution results. While the purpose is to provide a report to the user, this involves reading potentially sensitive information from the current session context.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection. It ingests untrusted data from the conversation history, code changes, and execution outputs (Ingestion points: SKILL.md, Section 2) and interpolates this data into an HTML template (templates/report.html) without explicit sanitization or boundary markers (Sanitization: Absent). This data is then rendered in a browser via the open command (Capability inventory: Bash open in SKILL.md, Section 4), creating a risk where malicious content previously introduced into the session (e.g., via a malicious file read or command output) could execute scripts in the context of the generated report.

session-reporter