The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local shell scripts (scripts/progress-snapshot.sh, scripts/render-html.sh, and scripts/plan-brief-open.sh) to gather data and generate the dashboard HTML. This relies on the existence and integrity of these scripts within the project's directory. Additionally, the project name is derived from the repository root directory name via git rev-parse and passed to shell commands, which could be a potential risk if the directory name contains shell metacharacters.
[PROMPT_INJECTION]: The skill processes untrusted content from the Plans.md file, which serves as a surface for indirect prompt injection.
Ingestion points: Data is read from Plans.md (specifically task titles and statuses like cc:WIP).
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when parsing the external file.
Capability inventory: The agent has access to Read, Write, and Bash tools.
Sanitization: There is no mention of sanitizing the content from Plans.md before it is processed by scripts or rendered into HTML, potentially allowing malicious content in task names to influence the generated dashboard.

harness-progress