The Agent Skills Directory

[PROMPT_INJECTION]: The skill defines automated workflows for issue triaging, documentation synchronization, and meeting note integration that ingest content from external sources. This creates a surface for indirect prompt injection, where malicious instructions hidden in a Jira issue's description or a Confluence page could attempt to influence the agent's automated actions.
Ingestion points: Content is ingested through tools like jira_get_issue (fetching summaries and descriptions) and confluence_get_page (fetching page content) as seen in references/common-workflows.md.
Boundary markers: The implementation lacks explicit delimiters or instructions to the agent to disregard instructions embedded within the ingested data.
Capability inventory: The skill provides broad write access to Atlassian APIs, including capabilities to create/update issues and modify wiki pages across its reference files.
Sanitization: An escapeHtml utility is provided, but it is used for formatting HTML storage payloads rather than sanitizing input data for secure prompt interpolation.
[EXTERNAL_DOWNLOADS]: The documentation provides configuration and installation steps for several external MCP server implementations from the community.
Packages: References the installation of @sooperset/mcp-atlassian (npm), mcp-atlassian (PyPI), and atlassian-mcp (npm) to enable the MCP integration features.

atlassian-mcp