The Agent Skills Directory

[SAFE]: The skill implements secure authentication using OAuth 2.1 with PKCE and API tokens. Credentials are stored in the system's secure keyring (e.g., macOS Keychain, Windows Credential Locker) using the keyring package, which prevents sensitive data from being stored in plain text or shell history.
[SAFE]: Network operations are directed to official and well-known Atlassian domains (atlassian.com, atlassian.net) for Jira, Confluence, and Atlassian's Model Context Protocol (MCP) services. This is standard behavior for an integration of this type.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Jira issues and Confluence pages, creating a potential surface for indirect prompt injection where instructions embedded in project content could target the agent.
Ingestion points: External data is retrieved via scripts/jira.py (issue summaries/descriptions) and scripts/confluence.py (page content).
Boundary markers: Absent. The skill does not currently wrap retrieved content in delimiters or include instructions for the agent to ignore embedded commands.
Capability inventory: The skill possesses capabilities for network communication with Atlassian APIs and local access to the system keyring.
Sanitization: The skill performs structural conversion from Atlassian Document Format (ADF) and HTML to plain text, which provides basic content filtering but does not neutralize linguistic instructions. Because this behavior is central to the skill's primary function, the risk is categorized as low and managed by the agent's core safety guardrails.

atlassian