The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute several local scripts bundled within the skill directory. These include dependency-mapper.sh for import analysis, render-spec.mjs for dashboard validation, and post_explore_summary.py for generating session summaries.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from the repository being explored, which constitutes a potential surface for indirect prompt injection if analyzed files contain malicious instructions.
Ingestion points: Codebase files are read via Grep, Glob, and Read tools during the exploration phases defined in SKILL.md and referenced rules like exploration-agents.md.
Boundary markers: While the subagent prompts (e.g., in exploration-agents.md) define technical scopes, they do not include explicit instructions to ignore instructions embedded within the codebase data.
Capability inventory: The skill has access to privileged tools including Bash, Write, and Agent, which could be misused if a subagent were to follow instructions embedded in codebase files.
Sanitization: No specific sanitization or filtering logic is applied to the codebase content before it is processed by the AI subagents.
[EXTERNAL_DOWNLOADS]: The post_explore_summary.py script references the yg-mcp-core library and mentions a private registry pypi.yonyon.ai. This is identified as a vendor-specific resource belonging to the author 'yonatangross' and is used for optional MCP-based summarization features.

explore