The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. It reads data from an external file (outline.yaml) and interpolates it directly into the prompt template for background subagents.
Ingestion points: Data is ingested from outline.yaml (Step 1) and fields.yaml (Step 3).
Boundary markers: The prompt template uses Markdown headers (e.g., ## Task, ## Field Definitions) but lacks specific guardrails or instructions to ignore potential commands embedded within the interpolated {item_related_info} variable.
Capability inventory: The skill uses the Task tool to spawn subagents, which in turn have access to the shell (Bash) to run Python scripts.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from outline.yaml before it is passed to subagents.
[COMMAND_EXECUTION]: The skill instructs background subagents to execute a local Python validation script (~/.claude/skills/research-outline/validate_json.py).
This script is expected to exist at a hardcoded path within the user's home directory based on the installation of the broader research skill suite.

research-deep