The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the task transcript and tool evidence to build a self-contained HTML report.
Ingestion points: Task transcript, real findings, paths, verified locations, metrics, and decisions gathered during the agent's task.
Boundary markers: The instructions do not specify the use of delimiters or boundary markers to isolate untrusted data within the HTML templates.
Capability inventory: The skill uses bash to run a preparation script, mkdir to create directories, and open to render the resulting HTML in a desktop browser.
Sanitization: No explicit HTML escaping or sanitization of the ingested task evidence is mentioned before it is interpolated into the report template.
[EXTERNAL_DOWNLOADS]: The prepare.sh script provides an installation command for the user to fetch the html-playground dependency from the author's repository (PaulRBerg/agent-skills). This is a reference to a vendor-owned resource.
[COMMAND_EXECUTION]: The skill executes a bundled shell script (prepare.sh) to validate the environment and uses the system open command to launch the generated HTML report in the default browser. The script includes slug validation to mitigate command injection risks.
[DYNAMIC_EXECUTION]: The skill generates a self-contained HTML file based on task evidence and templates, which is then executed (rendered) in a browser context by the open command.

html-debrief