The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection.
Ingestion points: The workflow utilizes web_fetch and github-mcp-server-get_file_contents to ingest data from arbitrary external sources, including third-party GitHub repositories like awesome-claude-code.
Boundary markers: The instructions do not define boundary markers or explicit safety guidelines to prevent the agent from following instructions potentially embedded within the fetched external content.
Capability inventory: The agent has the ability to interact with a session SQL database, generate structured markdown backlogs, and delegate tasks to subsequent skills such as skill-creator.
Sanitization: There is no mention of sanitization, filtering, or validation of the fetched data before it is interpolated into prompts or stored in the database.
[COMMAND_EXECUTION]: The skill uses dynamic SQL generation to manage candidates. Untrusted data from external sources is inserted into INSERT statements, which creates a risk of SQL injection within the local session database context if the source content is maliciously crafted.

ecosystem-intake