The Agent Skills Directory

[SAFE]: The skill's primary function involves reading project context files and interacting with the user to generate marketing hypotheses. These operations are confined to the local filesystem and do not involve network access, credential usage, or sensitive system interactions.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via the processing of external context files and user input.
Ingestion points: The skill reads company-specific data from claude-code-gtm/context/{company}_context.md and accepts conversational input from the user.
Boundary markers: The skill does not define specific delimiters to isolate external content from its own instructions.
Capability inventory: Available tools are limited to reading and writing markdown files. The skill does not possess capabilities for network requests, shell command execution, or dynamic code evaluation.
Sanitization: Content is processed through the agent's reasoning engine without explicit sanitization or filtering of input file contents.
Risk Assessment: The attack surface is minimal as the skill only outputs structured text to a markdown file and does not trigger high-risk capabilities.

hypothesis-building