The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates a resumable journey by reading and processing data from local files within the docs/ directory. This creates an indirect prompt injection surface where the agent's logic could be influenced if these files are manipulated by untrusted external processes.
Ingestion points: docs/IMPROVE-BUSINESS-PLAN.md, docs/STRATEGY.md, docs/OPERATIONS.md, docs/METRICS.md, docs/CUSTOMER.md, docs/OFFER.md (via Operating Rule 1).
Boundary markers: The instructions do not specify explicit delimiters or 'ignore' instructions for the content read from these files.
Capability inventory: The agent can write to the local file system and invoke other skills to perform tasks.
Sanitization: The skill relies on natural language processing of the file contents without technical sanitization layers.
[EXTERNAL_DOWNLOADS]: The instructions direct the agent to suggest the installation of additional skills from the author's official repository (wondelai/skills) using npx as a fallback mechanism if a required sub-skill is missing from the environment.
[COMMAND_EXECUTION]: The skill instructs the agent to provide the user with a specific command-line string (npx skills add ...) for manual execution. This requires user consent and action to modify the environment and is not performed autonomously by the agent.

improve-business