The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs automated file system operations including creating directories, writing files, and modifying existing project configuration files (CLAUDE.md). It also implements a discovery mechanism that searches ~/proyectos/ recursively (up to 2 levels) and parent directories to resolve project paths from simple names.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes content from external files that are conceptually outside the immediate project's trust boundary.
Ingestion points: The agent reads Markdown files from a shared/ directory, including files authored by other projects (to_{{project_local}}_*.md and from_{{proyecto_remoto}}_*.md).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the content of these messages as untrusted data or to ignore embedded instructions.
Capability inventory: The skill possesses the ability to read and write files within the project structure, modify CLAUDE.md, and perform path resolution/searches via the shell environment.
Sanitization: The skill does not perform any sanitization, escaping, or validation of the message content before presenting it to the agent context.

project-bridge