The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from Jira issues and Confluence pages, which constitutes an indirect prompt injection attack surface.\n
Ingestion points: Jira issue descriptions and Confluence page content are ingested in references/common-workflows.md through tools like jira_get_issue and confluence_get_page.\n
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded commands in the data interpolation examples.\n
Capability inventory: The skill allows for various write operations across its references, including jira_update_issue, jira_add_comment, confluence_create_page, and confluence_update_page.\n
Sanitization: A utility function escapeHtml is provided in references/common-workflows.md to sanitize data for the Confluence storage format.\n- [COMMAND_EXECUTION]: The skill provides configuration guidance for running shell-based MCP servers.\n
Evidence: SKILL.md and references/mcp-server-setup.md include command-line configurations for using npx and uvx to execute Atlassian MCP servers.\n- [EXTERNAL_DOWNLOADS]: The skill directs users to install external MCP server packages.\n
Evidence: References @anthropic/mcp-atlassian, mcp-atlassian (sooperset), and atlassian-mcp (xuanxt) as dependencies to be installed via standard package managers like npm, pip, or uv.

atlassian-mcp