The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data originating from project documentation and external user evidence such as support tickets and reviews.
Ingestion points: The agent reads state and content from files in the docs/ directory, including IMPROVE-APP-PLAN.md, CUSTOMER.md, DESIGN.md, POSITIONING.md, EXPERIMENTS.md, and PRODUCT.md.
Boundary markers: The operating rules instruct the agent to read artifacts before writing and to preserve existing content in the files when extending them.
Capability inventory: The skill is authorized to perform file write operations within the docs/ directory and to invoke sub-skills.
Sanitization: No specific sanitization or filtering logic is defined for external feedback or evidence data processed during analysis.
[COMMAND_EXECUTION]: The instructions include a fallback mechanism that suggests a shell command for the user to install missing components.
Evidence: The agent is instructed to suggest the command npx skills add wondelai/skills/<slug> --global if a required skill is not available.
[EXTERNAL_DOWNLOADS]: The skill facilitates the fetching of external resources from the author's repository to support the orchestration journey.
Evidence: The workflow references and suggests downloading sub-skills from the wondelai/skills/ namespace.

improve-app