The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs local file management using mkdir -p and cp commands to maintain wargame journals and prepare HTML dashboard templates. It also invokes external visualization tools such as mmdc (Mermaid) and dot (Graphviz) to render strategic diagrams based on the analyzed scenarios.
[EXTERNAL_DOWNLOADS]: It leverages WebSearch and WebFetch to gather relevant contextual information and intelligence from external sources to enrich the strategic analysis process.
[PROMPT_INJECTION]: The skill processes untrusted external data from user scenarios and web search results, which presents a surface for indirect prompt injection.
Ingestion points: Processes user-provided scenario text in SKILL.md and retrieves external intelligence through WebSearch in session-commands.md.
Boundary markers: Implements a structured 'Scenario Understanding' and 'Alignment Confirmation' phase where the agent synthesizes inputs and requires explicit user approval before proceeding with the analysis.
Capability inventory: Includes file system write access for session persistence in ~/.claude/wargames/, shell command execution for file management and diagram rendering, and the use of Playwright for capturing dashboard screenshots.
Sanitization: Uses structured JSON data interpolation for the HTML dashboard and employs an interactive confirmation loop to validate the scenario context before the wargame or analysis begins.

wargame