The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the ctx7 CLI to perform library lookups, fetch documentation, and manage agent instructions. This includes executing shell commands to configure environment variables and modify local agent configuration files such as .mcp.json and .claude/skills for various editors.
[EXTERNAL_DOWNLOADS]: Instructions guide the user to install the ctx7 utility via the NPM registry and to download additional agent skills from GitHub repositories using the ctx7 skills install command. These resources are fetched from well-known public registries and source control platforms.
[DATA_EXFILTRATION]: Documentation search queries and library resolution requests are transmitted to the Context7 API. The skill documentation includes a proactive security warning advising users to avoid including sensitive or personal data in their search terms.
[REMOTE_CODE_EXECUTION]: The tool supports the installation and execution of external instruction sets (skills) from GitHub repositories and the generation of new instruction files via an AI service, which are subsequently integrated into the agent's context.
[PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection by ingesting data from external documentation and remote skills while possessing the capability to execute commands and modify system files.
Ingestion points: External documentation fetched via ctx7 docs (references/docs.md) and third-party skills installed via ctx7 skills install (references/skills.md).
Boundary markers: The skill does not implement explicit delimiters or boundary warnings for the content returned from documentation queries.
Capability inventory: Shell command execution via the ctx7 CLI and file system writes for agent configuration as described in references/setup.md.
Sanitization: While users are warned not to provide sensitive inputs, there is no technical validation or sanitization of the content retrieved from the remote service.

context7-cli