The Agent Skills Directory

[PROMPT_INJECTION]: No patterns of instruction overriding, safety filter bypasses, or system prompt extraction were found in the skill text.\n- [COMMAND_EXECUTION]: The skill uses low-risk, read-only commands such as which acli, acli jira --action getIssue, and grep to gather context from the environment and Atlassian tools.\n- [EXTERNAL_DOWNLOADS]: Data retrieval is confined to official Atlassian endpoints via acli or MCP functions. No unauthorized remote script execution or external package installations are present.\n- [DATA_EXFILTRATION]: While the skill accesses business requirements and source code, it does so for internal architectural analysis. No patterns suggesting data transmission to unauthorized domains were detected.\n- [SAFE]: The skill performs its stated purpose using standard developer tools without requesting elevated privileges, attempting persistence, or employing obfuscation.

technical approach