The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions to install and execute additional packages from an external repository using npx skills add mattpocock/skills --skill <name>. This allows for the dynamic download and execution of arbitrary code from a non-vendor source at runtime.
[DATA_EXFILTRATION]: The skill instructions direct the agent to read .waypoint/local.yaml to extract the actor_id and other identifiers. This involves accessing local configuration files containing potentially sensitive identity metadata.
[PROMPT_INJECTION]: The skill processes external content such as feature specifications and technical designs while maintaining capabilities to write files and execute shell commands, creating a surface for indirect prompt injection.
Ingestion points: Accesses feature.md, .waypoint/local.yaml, and repository-level instructions.
Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for the ingested data.
Capability inventory: File system write access for publishing tasks and shell execution via the npx command.
Sanitization: The instructions do not specify any validation or sanitization for external content before processing.

task-planning