The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted email content which serves as a potential surface for indirect prompt injection attacks.
Ingestion points: External data is ingested through the agently-cli message +read, +list, and +search commands as described in SKILL.md.
Boundary markers: The skill contains explicit instructions to the agent to treat email body and subject content strictly as data and to ignore any embedded instructions.
Capability inventory: The agent has the ability to send, reply, forward, and delete emails, as well as download attachments to the local file system.
Sanitization: A mandatory two-stage confirmation mechanism is enforced for all destructive or outgoing operations (+send, +reply, +forward, +trash), ensuring that the agent cannot execute these actions without explicit human approval after reviewing the operation summary.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @tencent-qqmail/agently-cli package and uses npx to add the skill content.
Evidence: npm install -g @tencent-qqmail/agently-cli and npx skills add Tencent/AgentlyMail -g -y.
These resources originate from the official infrastructure of the vendor (Tencent) and are standard for the skill's functionality.
[COMMAND_EXECUTION]: The skill's primary functionality is delivered through the execution of shell commands using the agently-cli tool.
Evidence: Use of various commands such as agently-cli message +send and agently-cli attachment +download to manage email resources.

agently-mail