The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources that could influence the agent's behavior.
Ingestion points: Data is ingested from HubSpot CRM records via hubspot activities list and hubspot activities calls transcript get in SKILL.md.
Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore embedded commands within the CRM data.
Capability inventory: The skill uses hubspot CLI commands to read and write CRM data and uses shell piping to manipulate data with jq.
Sanitization: The skill does not perform sanitization or filtering of the content retrieved from transcripts or activity bodies before presenting it to the agent context.
[COMMAND_EXECUTION]: The skill uses standard shell utilities and a vendor-specific CLI tool to perform its tasks.
Evidence: Uses hubspot CLI, jq, date, and echo in several bash pipelines within SKILL.md to automate CRM data retrieval and formatting.

communication-history