The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources such as Plane tickets, Bloodbank events, and human directives as specified in references/task-execution.md and workflows/task-intake.md. This design pattern introduces a surface for indirect prompt injection where malicious instructions could be embedded in ticket descriptions or event payloads.
Ingestion points: External task metadata (Plane tickets), event streams (Bloodbank), and manual directives.
Boundary markers: Absent. The instructions do not specify using delimiters or "ignore instructions" wrappers when reading task content.
Capability inventory: The agent is instructed to perform file system modifications, code generation, and execute deployment commands (Docker, Cloudflare CLI) based on these inputs.
Sanitization: Not identified. The skill relies on the agent to interpret the input directly.
[COMMAND_EXECUTION]: The skill documentation (references/infrastructure-deployment.md, workflows/project-bootstrap-pjangler.md) contains numerous shell commands for Docker management, Cloudflare tunnel configuration, and project initialization using the vendor-provided pjangler tool. These commands are integral to the skill's stated purpose of system orchestration.
[EXTERNAL_DOWNLOADS]: The workflows describe fetching and installing dependencies for the pjangler tool from its local source directory and utilizing well-known external services like Cloudflare and Traefik. These operations are consistent with standard development and deployment practices within the specified ecosystem.

33god