From omni
Orchestrates multi-module SRP (Single Responsibility Principle) compliance analysis across a codebase. Supports full-scan and git-diff-based incremental mode for CI gating. Identifies modules, runs concurrent SRP checks, then aggregates results into a summary report.
How this skill is triggered — by the user, by Claude, or both
Slash command
/omni:ai-friendly-component-srp-orchestrateThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
以代码库为粒度,完成模块一致性检测。支持**全量模式**和**增量模式**。
CHANGELOG.mdCHANGES-DETAIL.mdDESIGN.mdREADME-INCREMENTAL.mdevals/eval-aggregate.pyevals/eval-gate-check.pyscripts/adapt-step02.pyscripts/aggregate.pyscripts/gate-check.pyscripts/identify-changed-modules.shscripts/test-gate-fail.shscripts/test-incremental.shworkflow/step00.5-identify-changed-modules.mdworkflow/step01-identify-modules.mdworkflow/step02-analyze-modules.mdworkflow/step03-aggregate-results.md以代码库为粒度,完成模块一致性检测。支持全量模式和增量模式。
origin/master)HEAD)false)true,自动检测 git ls-files --others)false)| Step | 职责 | 执行者 | 文档 | 输入 | 输出 |
|---|---|---|---|---|---|
| 01 | 识别所有模块 | SubAgent → ai-friendly-module-identifier | step01 | project_path | state/modules.json |
| 02 | 并发 SRP 分析(每批5个) | Main Agent → 多 SubAgent → ai-friendly-arch-guard-module-single-responsibility | step02 | state/modules.json | state/step02-analyze-modules/*.json |
| 03 | 聚合汇总 | 脚本 scripts/aggregate.py | step03 | state/step02-analyze-modules/*.json | output/summary.json |
| Step | 职责 | 执行者 | 文档 | 输入 | 输出 |
|---|---|---|---|---|---|
| 01 | 识别所有模块 | SubAgent → ai-friendly-module-identifier | step01 | project_path | state/modules.json |
| 02 | 识别变更模块(含 Orphan 文件追踪) | 脚本 scripts/identify-changed-modules.sh | step02 | state/modules.json + base_commit + target_commit | state/changed-modules.json(含 orphan_files 字段) |
| 03 | 并发 SRP 分析(每批5个) | Main Agent → 多 SubAgent → ai-friendly-arch-guard-module-single-responsibility | step03 | state/changed-modules.json | state/step02-analyze-modules/*.json |
| 04 | 聚合汇总 | 脚本 scripts/aggregate.py | step04 | state/step02-analyze-modules/*.json + state/changed-modules.json | output/srp/summary.json |
| 05 | 门禁判定(可选) | 脚本 scripts/gate-check.py | - | output/srp/summary.json + .gate-config.json | output/gate-result.json |
state/ 目录;各 SubAgent 只写自己负责的文件,返回极简 JSONstate/
modules.json # step01 产物:全量模块清单
changed-modules.json # step0.5 产物:变更模块清单(增量模式)
step02-analyze-modules/ # step02 产物:每模块 SRP 分析
{module_name}.json
output/
srp/
summary.json # step03 产物:最终汇总报告(aia_metric_fact 格式)
gate-result.json # step04 产物:门禁判定结果(可选)
.gate-config.json # 门禁阈值配置(可选)
output/summary.json(aia_metric_fact 格式)identity_info.skill_id:"ai-friendly-component-srp-orchestrate"identity_info.arch_dimension:"结构可导航性"execution_ctx.skill_version:技能版本号execution_ctx.scan_mode:扫描模式("full" 或 "increment")execution_ctx.execute_status:执行状态("success" 或 "failed")execution_ctx.start_time / end_time / duration_ms:执行时间信息core_metrics.total_score:综合得分(0~100)core_metrics.confidence_score:置信度得分(0~1)core_metrics.total_violation_count:总违规数core_metrics.p0_violation_count:P0 级别违规数core_metrics.p1_violation_count:P1 级别违规数evaluation_details.score_detail:各子维度得分明细(目录单一性、模块内聚、文件单一性)evaluation_details.score_distribution:得分分档分布(excellent/good/medium/poor)violation_records.level_summary:违规级别汇总(P0/P1)violation_records.violation_infos:违规详情列表(最多 50 条)scan_statistics.total_units.modules:扫描模块总数scan_statistics.violation_units.modules:存在违规的模块数scan_statistics.valid_units.modules:无违规模块数modules:每个模块的摘要(扩展字段,非 aia_metric_fact 标准字段)state/changed-modules.json(增量模式输出,step0.5 产物)mode:扫描模式("incremental")base_commit / target_commit:对比区间modules:按架构层分类的变更模块列表,每个模块含 changed_filesorphan_files:未匹配到任何模块的变更文件(新增模块、删除文件残留)statistics.orphan_files_count:orphan_files 数量output/gate-result.json(启用门禁时)gate_passed:是否通过门禁(true/false)thresholds:使用的阈值配置actual_values:实际度量值violations:未通过的检查项调用技能 ai-friendly-component-srp-orchestrate
参数:
project_path: /path/to/repo
调用技能 ai-friendly-component-srp-orchestrate
参数:
project_path: /path/to/repo
base_commit: origin/master
enable_gate: true
# 跳过 untracked 新增文件,仅分析已 commit 的 diff
参数追加:detect_untracked: false
# 同时检测已 staged 但未 commit 的变更
参数追加:detect_staged: true
编辑 .gate-config.json 自定义阈值:
{
"min_avg_score": 0.7,
"max_avg_violation_count": 5,
"min_confidence": 0.6
}
npx claudepluginhub zte-aicloud/co-omnispec --plugin omni分析指定模块目录的单一职责原则遵守情况,生成三维度得分与违规报告的JSON。适用于检查模块内聚性、检测职责混杂问题。
Analyzes codebase architecture via multi-agent specialists on structure, coupling, integration, error handling, security; verifies findings, reports strengths and flaws with evidence.
Refactors existing code for guideline compliance and testability using 4 parallel analysis agents. Use after /optimus:init to align code or before /optimus:unit-test to remove testability barriers.