菜单
Run a full IRIS instance health check (iris_health_check) and interpret every non-ok finding, naming the fixing tool for each one.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Review and promote configuration/code drift from a source IRIS environment to a target using the review-before-write iris_env_diff -> iris_env_promote workflow (scoped diff, plan, an explicit user-selected step allowlist, confirmed execute, re-diff verify). Never acts on onlyInTarget warnings -- never deletes a target-only item, in any action.
Resend failed Interoperability messages for a config item using the dry-run-first iris_message_resend workflow — preview the match count before executing, then verify the new headers via iris_production_messages. States the duplication hazard and that resend/resendFiltered are governance-default-disabled writes.
Audit the IRIS instance's security posture: users (default passwords, %All holders), roles, service authentication settings, SSL/TLS configs, and instance auditing status.
Deploy an ObjectScript class or package to IRIS (iris_doc_load, glob-path form), fix compile errors, then run its unit tests (iris_execute_tests) with a total-count check.
Diagnose a slow SQL query using iris_sql_analyze (explain, indexUsage, stats) and recommend a fix — never auto-applies one.
A concise (<=300 word) pre-write checklist distilling the project's ObjectScript conventions ($$$ macros, Quit-in-try/catch, %OnNew/initvalue, no-underscore names, storage sections untouchable, deploy/test discipline).
| name | check-system-health |
| description | Run a full IRIS instance health check (iris_health_check) and interpret every non-ok finding, naming the fixing tool for each one. |
No server profile specified — omit server (or pass "") to use the default server on every tool call below.
iris_health_check (optionally with an areas filter; omit or pass [] for all 9 areas) to get one composite verdict ('healthy' | 'warning' | 'critical') plus a per-area finding explaining WHY.verdict first, then walk every finding whose level is 'warning', 'critical', or 'error' — skip 'ok' and 'notApplicable' findings, they need no action.explanation may already name a fixing tool. Otherwise use this mapping:
iris_journal_info for detail; free disk space, or run a configured purge task via iris_task_run.iris_database_check for the named worst database. To raise maxSize/expansionSize, use iris_database_manage action 'modify' — note this tool is on the admin MCP server, NOT this ops server, so switch to the admin server for that step (or free disk space instead).iris_license_info for detail; reduce active connections or contact licensing.iris_locks_list to find the largest lock holders; consider iris_process_manage to investigate the offending process.iris_metrics_alerts for the active alert detail.iris_journal_info, iris_mirror_status).iris_database_manage modify, iris_task_run) without the user's explicit confirmation.