The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core function of reading and acting upon data from external, potentially untrusted sources like Jira issue descriptions and Confluence page content.\n
Ingestion points: Untrusted data enters the agent's context through tools such as jira_get_issue, jira_get_issue_comments, confluence_get_page, and confluence_get_page_comments as described in SKILL.md.\n
Boundary markers (absent): While the skill includes instructions for the agent to wait for human confirmation (Constitution XIV), it lacks technical boundary markers or specific delimiters to isolate external data from the system's own instructions.\n
Capability inventory: The skill provides access to several write-enabled tools including jira_create_issue, jira_update_issue, jira_transition_issue, confluence_create_page, and confluence_update_page, which could be abused if the agent inadvertently follows instructions embedded in an incident report or wiki page.\n
Sanitization (absent): There is no mention of automated sanitization, filtering, or validation of the text content retrieved from the Atlassian tools before it is processed by the agent.

atlassian-itsm