The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow that ingests external, potentially untrusted data from user requirements and project metadata, which could contain hidden instructions intended to override agent logic. * Ingestion points: Processes user stories, requirement specifications, bug reports, and project files like package.json or angular.json during Phase 1 and Phase 2. * Boundary markers: The skill does not provide specific instructions to use delimiters or warnings to ignore instructions embedded within the requirement text. * Capability inventory: The skill has access to file writing and editing (Write, Edit), shell command execution (Bash), and the ability to invoke other specialized skills via a Skill tool. * Sanitization: There are no explicit instructions for the sanitization or validation of requirement inputs.
[DATA_EXFILTRATION]: The skill implements a security best practice by explicitly instructing the agent to refuse the handling of sensitive credential files. * Evidence: The skill's Reference.md instructions mandate that the agent must refuse to stage secret-like files, specifically naming .env, *.pem, and credentials files.

angular-dev