The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is designed to read sensitive local file paths to provide context for strategic analysis. It specifically accesses 'me.md', which typically contains personal developer information, along with project files containing financial metrics and business data.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted project data (e.g., README.md, git logs, and configuration files) and interpolates it into prompts for secondary agents. This creates a surface where malicious instructions hidden in these files could influence sub-agent behavior.
Ingestion points: CLAUDE.md, AGENTS.md, README.md, me.md, .claude/rules/, git logs, and financial data files.
Boundary markers: No explicit delimiters or instructions are utilized to separate project data from the persona prompts in the sub-agent tasks.
Capability inventory: The skill uses the Task tool for sub-agent execution, performs local file system read/write operations, and executes git commands.
Sanitization: No sanitization or validation of external content is performed before interpolation into prompts.
[COMMAND_EXECUTION]: The skill instructions require the agent to execute shell commands to gather context, specifically 'git log --oneline -10', to inform the strategic analysis.
[EXTERNAL_DOWNLOADS]: The skill references an external GitHub repository (https://github.com/serejaris/visual-grammar) for generating visual assets. This resource is maintained by the skill's author.

ceo-council