The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill utilizes npx to download and execute the @useclawlink/cli package, which is the primary mechanism for interacting with Dynamics 365 actions.
[EXTERNAL_DOWNLOADS]: The skill fetches the official @useclawlink/cli package from the NPM registry to provide the necessary integration tools.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for user login, connecting services, and running Dynamics 365 actions (e.g., creating accounts or leads).
[CREDENTIALS_UNSAFE]: The skill manages authentication by storing credentials in ~/.clawlink/credentials.json following the execution of the login command. This is standard behavior for the vendor's command-line interface.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from Dynamics 365, creating a potential surface for indirect prompt injection if record content contains malicious instructions.
Ingestion points: Data is ingested through CLI actions that read leads, invoices, and other CRM entities (SKILL.md).
Boundary markers: The skill explicitly instructs the agent to "confirm with the user before any write," providing a safety checkpoint (SKILL.md).
Capability inventory: The agent can perform write operations including creating and updating accounts, cases, and opportunities via shell commands (SKILL.md).
Sanitization: There is no mention of explicit sanitization or filtering of the content retrieved from Dynamics 365 records.

dynamics-365