The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests untrusted data from issue trackers (titles, bodies, and comments) and code conflict hunks. This data is provided to subagents (analyst, resolver, critic) to automate merge planning and conflict resolution. Ingestion points include 'gh issue view' output and local source files. The skill lacks explicit boundary markers or sanitization for this data, creating an indirect prompt injection surface. Capabilities include shell execution and file system writes.
[COMMAND_EXECUTION]: The skill frequently executes system commands using 'git', 'gh', 'curl', and 'oha' via the child_process module. These operations are essential for its core functionality of repository management, API interaction, and integration testing.
[DATA_EXFILTRATION]: The skill accesses local configuration at '~/.issue/settings.json' and sensitive environment variables such as 'JIRA_API_TOKEN' and 'CONFLUENCE_API_TOKEN'. This data is used for legitimate authentication with Jira and Confluence services to which the skill transmits report data.

issue-merge