The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It reads content from requirements.md and various task-*.md files and incorporates this data into an execution-handoff.md file. This generated file includes a specific section titled "Prompt for implementer" which is intended to be used as instructions by a downstream agent. If an attacker or malicious process modifies the source task files, they could inject instructions that influence the behavior of the agent performing the implementation phase.
Ingestion points: The skill reads data from {product_root}/docs/versions/{version_san}/requirements.md and files matching task-*.md in the same directory.
Boundary markers: The data is structured using Markdown tables in the execution-handoff.template.md, which provides visual structure but does not prevent an LLM from interpreting instructions embedded within the text fields.
Capability inventory: The skill performs file system read and write operations and generates instructional content meant to guide the workflow of other agents.
Sanitization: The skill instructions do not specify any sanitization, filtering, or escaping of the ingested content before it is placed into the output template.

execution-handoff-generator