The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill delegates its primary logic to a remote file located at https://github.com/datopian/portaljs/blob/main/.claude/commands/portaljs-architect.md, identifying it as the single source of truth for the workflow. This introduces a runtime dependency on external content.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it incorporates untrusted user requirements and data from local files into its decision-making process.
Ingestion points: Data enters the context via the $ARGUMENTS parameter and through metadata or contents of local files inspected by the Bash tools.
Boundary markers: There are no specific instructions or delimiters provided to the agent to distinguish between valid data and potentially malicious instructions embedded within the user-provided data or local files.
Capability inventory: The skill can write files (./ARCHITECTURE.md), execute shell commands for file analysis, and initiate other setup skills.
Sanitization: No validation or sanitization routines are defined to handle external content before it is processed by the agent's logic.
[COMMAND_EXECUTION]: The skill suggests the use of command-line tools like perl for manual modification of project files (perl -pi -e) during troubleshooting and uses system tools to inspect file metadata.

portaljs-architect