The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data exfiltration were detected. The skill's behavior aligns with its documented purpose of generating marketing copy and scaffolding project files.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it ingests data from local project files (PRDs, READMEs) and has the capability to modify the filesystem or execute shell commands.
Ingestion points: Product information is read from PRD, README, or research.md (Step 3).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are used for the ingested data.
Capability inventory: The skill utilizes Write, Edit, and Bash tools, which allow for filesystem modification.
Sanitization: The skill performs structured extraction of specific fields (e.g., "Problem", "Solution", "Features"), which implicitly sanitizes the input by ignoring irrelevant or potentially malicious instruction blocks.
[COMMAND_EXECUTION]: The skill requests the Bash tool to detect project frameworks (e.g., Astro or Next.js) by checking for configuration files. This is a legitimate use of the tool for environment detection and scaffolding.
[EXTERNAL_DOWNLOADS]: The skill utilizes an MCP tool (mcp__solograph__web_search) to perform web searches for competitor analysis. This is an expected part of the research workflow and does not involve downloading or executing untrusted code.

solo-landing-gen