The Agent Skills Directory

[Category 2: Data Exposure & Exfiltration] (SAFE): The audit capability and the score_content.py script read local files to perform analysis. However, the script lacks any network capabilities (e.g., requests, urllib, socket), meaning data cannot be exfiltrated. Output is restricted to standard output.
[Category 4: Unverifiable Dependencies & Remote Code Execution] (SAFE): The skill does not install external packages or download remote scripts. The Python script relies entirely on the standard library.
[Category 8: Indirect Prompt Injection] (LOW): The skill has an attack surface for indirect prompt injection because it processes untrusted data from the web (top-ranking SERP pages) and local files.
Ingestion points: External content via search_web in SKILL.md and local file reading in scripts/score_content.py.
Boundary markers: Absent; there are no specific delimiters or instructions to ignore embedded commands in the ingested data.
Capability inventory: Web searching, local file reading, and executing a local Python script for text analysis.
Sanitization: The clean_text function in scripts/score_content.py provides basic sanitization by stripping code blocks, HTML/JSX tags, and frontmatter before processing text content.

surfer-seo-optimizer