The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands such as curl for retrieving documentation and relies on the neonctl CLI for project and database management. Evidence found in references/neon-cli.md and SKILL.md.
[DATA_EXFILTRATION]: The skill instructs the agent to read and update project .env files to configure database connection strings. This level of access is restricted to the intended primary purpose of database setup. Evidence found in references/getting-started.md.
[EXTERNAL_DOWNLOADS]: The skill fetches documentation indices and markdown content from neon.com. These downloads originate from a well-known technology service and are essential for providing the agent with current reference material. Evidence found in references/referencing-docs.md.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection vulnerabilities because it fetches and processes external content from neon.com without instructing the agent to use boundary markers or safety warnings.
Ingestion points: Documentation fetched via curl from https://neon.com/docs/<path> as described in SKILL.md and references/referencing-docs.md.
Boundary markers: Absent; fetched content is processed directly without delimiters.
Capability inventory: Shell command execution via neonctl, file system modifications (e.g., .env), and SQL execution via the Neon MCP server.
Sanitization: No instructions are provided to the agent to sanitize or escape instructions that may be embedded in the retrieved documentation.

neon-postgres