The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the compose-agentsmd command before responding to any user message as a 'session gate' in AGENTS.md. Additionally, the .mwt/config.toml file defines a verification command that executes npx markdownlint-cli and compose-agentsmd.
[EXTERNAL_DOWNLOADS]: The AGENTS.md file contains instructions to install the compose-agentsmd tool globally via npm install -g if it is missing. The verification script in .mwt/config.toml also uses npx --yes to download and execute markdownlint-cli at runtime.
[PROMPT_INJECTION]: The skill defines a process that is susceptible to indirect prompt injection by relying on external data sources.
Ingestion points: As specified in SKILL.md, the agent must use documentation, specifications, CLI help, and API descriptions as the source of truth for contract verification.
Boundary markers: The instructions lack any requirement for delimiters or warnings to ignore instructions found within these external documents.
Capability inventory: The agent environment allows for the execution of shell commands through the identified verification and rule management tools (compose-agentsmd, npx).
Sanitization: There is no requirement for the agent to sanitize or validate the external content before processing it.

quality-workflow