The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a shell script in Step 3 that uses the find command to dynamically search for a reference file (content-writing.md) within various hidden plugin and skill directories (e.g., ~/.claude/plugins, ~/.claude/skills). While intended for local path discovery, this involves active shell command execution.
[PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection because its primary function is to ingest untrusted data from external URLs (via firecrawl/web tools) or local files to perform content audits and rewrites. Malicious instructions embedded in these external sources could attempt to influence the agent's behavior.
Ingestion points: Users provide a "URL to improve" or "file path" as the primary input.
Boundary markers: The instructions do not define clear boundary markers or instructions for the agent to ignore potentially malicious directions within the ingested content.
Capability inventory: The skill executes shell commands (find) and has the capability to generate code-like structures (JSON-LD) and read arbitrary local files if provided as paths.
Sanitization: No sanitization or validation of the external content is described before it is processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill's workflow explicitly instructs the agent to use external crawling tools like firecrawl or WebSearch to fetch content from remote URLs provided by the user.

content-writer