The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the Plain platform, such as customer emails, chat messages, and Slack replies, which are ingested via the thread timeline and thread search commands.
Ingestion points: scripts/plain-api.sh (customer, thread, and timeline retrieval functions).
Boundary markers: Absent. There are no instructions or delimiters provided to help the agent distinguish between its system instructions and the content of the support messages it reads.
Capability inventory: The skill has access to the Bash tool and can execute scripts/plain-api.sh, which performs file system reads and network operations.
Sanitization: JSON payloads are escaped using jq for API requests, but the text content returned from the API is processed by the agent without specific sanitization.
[COMMAND_EXECUTION]: The skill provides a comprehensive bash script (scripts/plain-api.sh) that uses curl and jq to interact with the Plain API. The script is well-implemented with safe JSON construction techniques, although it does perform file reads using cat when the agent provides a file path for notes or articles.
[DATA_EXFILTRATION]: The skill communicates with core-api.uk.plain.com and app.plain.com. These are the official endpoints for the Plain support platform, making the data transmission consistent with the skill's primary purpose.

plain-support