aiskillstore/marketplace / skills/a-genno/child-agent-ops
child-agent-ops
Verified子エージェントの起動・指示・報告・安全運用を標準化するスキル。codex でのタスク分解、確認判断、チェックリスト、テンプレートを提供し、リーダー/実装/レビュー各エージェントが迷わず動けるようにする。
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 4.8 KB |
| skill-report.json | config | 14.1 KB |
skill-report.json
View on GitHub{
"schema_version": "2.0",
"meta": {
"generated_at": "2026-07-07T19:17:14.312Z",
"slug": "a-genno-child-agent-ops",
"source_url": "https://github.com/A-Genno/Polaris/tree/master/documents/skills/child-agent-ops",
"source_ref": "master",
"model": "codex",
"analysis_version": "3.0.0",
"source_type": "community",
"content_hash": "71256186cd27bc0664e252f42adbcd21e9f42fd10e5a9fbbaccd90feca66358e",
"tree_hash": "d3204c86b82e2c12e3223cc61594599f7771a505726a99abaf89bd9a1b41f209",
"provenance": {
"model_effective": "codex",
"fallback_chain": [
{
"agent": "codex",
"outcome": "success"
}
],
"analysis_version": "3.0.0"
}
},
"skill": {
"name": "child-agent-ops",
"description": "Standardizes how to launch, assign, monitor, and report child agents for Codex workflows.",
"summary": "Provides a repeatable operating process for coordinated child-agent work.",
"icon": "📦",
"version": "1.0.1",
"author": "A-Genno",
"license": "MIT",
"category": "productivity",
"tags": [
"multi-agent",
"codex",
"task-management",
"workflow",
"code-review"
],
"supported_tools": [
"claude",
"codex",
"claude-code"
],
"risk_factors": [
"external_commands"
]
},
"security_audit": {
"risk_level": "safe",
"is_blocked": false,
"safe_to_publish": true,
"analysis_status": "ok",
"summary": "The static external-command alerts are Markdown backticks around the word codex and fenced report templates, not executable code. The high-entropy heuristic is a false positive caused by a plain Markdown file containing Japanese text. I found no prompt injection, data exfiltration, or malicious operational intent in SKILL.md.",
"remediation": [],
"risk_factor_evidence": [
{
"factor": "external_commands",
"evidence": [
{
"file": "SKILL.md",
"line_start": 10,
"line_end": 10
},
{
"file": "SKILL.md",
"line_start": 25,
"line_end": 25
},
{
"file": "SKILL.md",
"line_start": 34,
"line_end": 34
},
{
"file": "SKILL.md",
"line_start": 48,
"line_end": 60
},
{
"file": "SKILL.md",
"line_start": 60,
"line_end": 62
},
{
"file": "SKILL.md",
"line_start": 62,
"line_end": 69
},
{
"file": "SKILL.md",
"line_start": 69,
"line_end": 72
}
]
}
],
"static_findings": [
{
"id": "external_commands:SKILL.md:10:ruby-shell-backtick-execution",
"category": "external_commands",
"severity": "medium",
"pattern": "Ruby/shell backtick execution",
"file": "SKILL.md",
"line_start": 10,
"line_end": 10,
"snippet": "2. **`codex` で起動ログを残す**(種別・担当範囲・完了条件・想定時間)。依頼メッセージとセットで記録。"
},
{
"id": "external_commands:SKILL.md:25:ruby-shell-backtick-execution",
"category": "external_commands",
"severity": "medium",
"pattern": "Ruby/shell backtick execution",
"file": "SKILL.md",
"line_start": 25,
"line_end": 25,
"snippet": "2. 自律実行可能な単位に分割し、必要な専門性ごとに子エージェントを `codex` で起動(範囲・完了条件・想定時間を記録)。"
},
{
"id": "external_commands:SKILL.md:34:ruby-shell-backtick-execution",
"category": "external_commands",
"severity": "medium",
"pattern": "Ruby/shell backtick execution",
"file": "SKILL.md",
"line_start": 34,
"line_end": 34,
"snippet": "- **開始前**: 目的/完了条件/想定時間を決め、依頼メッセージと `codex` 起動ログを残す。"
},
{
"id": "external_commands:SKILL.md:48:ruby-shell-backtick-execution",
"category": "external_commands",
"severity": "medium",
"pattern": "Ruby/shell backtick execution",
"file": "SKILL.md",
"line_start": 48,
"line_end": 60,
"snippet": "```md"
},
{
"id": "external_commands:SKILL.md:60:ruby-shell-backtick-execution",
"category": "external_commands",
"severity": "medium",
"pattern": "Ruby/shell backtick execution",
"file": "SKILL.md",
"line_start": 60,
"line_end": 62,
"snippet": "```"
},
{
"id": "external_commands:SKILL.md:62:ruby-shell-backtick-execution",
"category": "external_commands",
"severity": "medium",
"pattern": "Ruby/shell backtick execution",
"file": "SKILL.md",
"line_start": 62,
"line_end": 69,
"snippet": "```md"
},
{
"id": "external_commands:SKILL.md:69:ruby-shell-backtick-execution",
"category": "external_commands",
"severity": "medium",
"pattern": "Ruby/shell backtick execution",
"file": "SKILL.md",
"line_start": 69,
"line_end": 72,
"snippet": "```"
},
{
"id": "obfuscation:SKILL.md:1:heuristic-high-file-entropy-7-28-bits-possible-b",
"category": "obfuscation",
"severity": "high",
"pattern": "[HEURISTIC] High file entropy (7.28 bits) - possible binary/encrypted content",
"file": "SKILL.md",
"line_start": 1,
"line_end": 1,
"snippet": "File: SKILL.md"
}
],
"finding_verdicts": [
{
"id": "external_commands:SKILL.md:10:ruby-shell-backtick-execution",
"verdict": "false_positive",
"confidence": 0.95,
"reason": "Line 10 uses Markdown inline code formatting for the word codex. It contains no shell execution syntax, arguments, or user-controlled command construction."
},
{
"id": "external_commands:SKILL.md:25:ruby-shell-backtick-execution",
"verdict": "false_positive",
"confidence": 0.94,
"reason": "Line 25 describes recording a Codex launch in prose. The backticks are Markdown formatting and do not execute a command."
},
{
"id": "external_commands:SKILL.md:34:ruby-shell-backtick-execution",
"verdict": "false_positive",
"confidence": 0.95,
"reason": "Line 34 references a codex startup log using inline Markdown. There is no script, command substitution, or unsafe input handling."
},
{
"id": "external_commands:SKILL.md:48:ruby-shell-backtick-execution",
"verdict": "false_positive",
"confidence": 0.97,
"reason": "Lines 48-60 are a fenced Markdown report template. The triple backticks delimit documentation content and are not Ruby or shell execution."
},
{
"id": "external_commands:SKILL.md:60:ruby-shell-backtick-execution",
"verdict": "false_positive",
"confidence": 0.97,
"reason": "Lines 60-62 only close one Markdown fence and open another template section. This is documentation formatting, not executable code."
},
{
"id": "external_commands:SKILL.md:62:ruby-shell-backtick-execution",
"verdict": "false_positive",
"confidence": 0.97,
"reason": "Lines 62-69 are a fenced Markdown child-agent launch log template. No command execution or shell interpolation is present."
},
{
"id": "external_commands:SKILL.md:69:ruby-shell-backtick-execution",
"verdict": "false_positive",
"confidence": 0.96,
"reason": "Lines 69-72 include the closing Markdown fence and a prose command policy. The finding is caused by formatting backticks, not an executable shell construct."
},
{
"id": "obfuscation:SKILL.md:1:heuristic-high-file-entropy-7-28-bits-possible-b",
"verdict": "false_positive",
"confidence": 0.93,
"reason": "SKILL.md is readable Markdown with Japanese prose and templates. I found no binary payload, encrypted blob, encoded script, or obfuscated instruction content."
}
],
"semantic_findings": [],
"critical_findings": [],
"high_findings": [],
"medium_findings": [],
"low_findings": [],
"dangerous_patterns": [],
"files_scanned": 1,
"total_lines": 80,
"audit_model": "codex",
"audited_at": "2026-07-07T19:17:14.312Z",
"agent_auto_install_policy": "allowed",
"manual_install_policy": "allowed"
},
"content": {
"user_title": "Coordinate Child Agent Operations",
"value_statement": "Parallel AI work fails when roles, checkpoints, and reports are unclear. This skill gives Claude, Codex, and Claude Code teams a repeatable child-agent operating process.",
"seo_keywords": [
"Claude",
"Codex",
"Claude Code",
"child agent operations",
"multi-agent workflow",
"AI task delegation",
"Codex review agent",
"agent reporting templates",
"parallel coding workflow",
"agent safety checklist"
],
"actual_capabilities": [
"Defines leader, implementation, and review agent responsibilities.",
"Provides a standard workflow for splitting and assigning child-agent tasks.",
"Lists approval checkpoints for security, billing, deletion, and compatibility risks.",
"Supplies report templates for task status and child-agent launch logs.",
"Requires review-agent involvement when code is changed or created."
],
"limitations": [
"Does not execute commands or create child agents by itself.",
"Does not provide an orchestration API or automation framework.",
"Assumes users already understand the supported AI tools.",
"Does not replace human approval for high-risk changes."
],
"use_cases": [
{
"title": "Coordinate Parallel Implementation",
"description": "Split an engineering task into clear child-agent assignments with completion criteria and checkpoints.",
"target_user": "Engineering lead"
},
{
"title": "Standardize Agent Reporting",
"description": "Use consistent status reports so blockers, risks, and next actions are easy to track.",
"target_user": "Project coordinator"
},
{
"title": "Add Review Coverage",
"description": "Trigger review-agent work when code changes need independent assessment before completion.",
"target_user": "Developer"
}
],
"prompt_templates": [
{
"title": "Create a Simple Assignment",
"scenario": "Beginner task delegation",
"prompt": "Use child-agent-ops to define one child-agent assignment for this task. Include purpose, scope, completion criteria, and expected duration."
},
{
"title": "Split Work Across Agents",
"scenario": "Intermediate parallel planning",
"prompt": "Use child-agent-ops to split this project into leader, implementation, and review responsibilities. Include checkpoints and reporting expectations."
},
{
"title": "Review Operational Risk",
"scenario": "Advanced safety review",
"prompt": "Apply child-agent-ops to identify which parts of this plan require leader approval before a child agent continues."
},
{
"title": "Run a Full Agent Operation",
"scenario": "Advanced multi-agent coordination",
"prompt": "Use child-agent-ops to produce a complete operating plan for multiple child agents, including launch logs, checkpoint timing, escalation rules, and final reports."
}
],
"output_examples": [
{
"input": "Plan child-agent work for a small refactor.",
"output": "A leader plan with scoped implementation and review assignments, clear completion criteria, and checkpoint notes."
},
{
"input": "Report progress from a delegated investigation.",
"output": "A concise task report covering purpose, current status, open issues, risks, and next actions."
},
{
"input": "Decide whether a child agent can proceed without approval.",
"output": "A decision note that separates autonomous work from changes needing leader review, such as security or compatibility impact."
}
],
"best_practices": [
"Define completion criteria before launching any child agent.",
"Escalate unclear security, billing, deletion, or compatibility risks early.",
"Record final outputs, remaining risks, and next actions in one place."
],
"anti_patterns": [
"Launching child agents without a defined owner or scope.",
"Letting parallel agents modify overlapping files without coordination.",
"Skipping review-agent checks after code changes."
],
"faq": [
{
"question": "Does this skill create child agents automatically?",
"answer": "No. It provides an operating process and templates for users who run supported tools."
},
{
"question": "Which tools does it support?",
"answer": "The report lists Claude, Codex, and Claude Code as supported tools."
},
{
"question": "Is this only for software projects?",
"answer": "It is written for coding workflows, but the reporting structure can fit other technical tasks."
},
{
"question": "When should a leader approve work?",
"answer": "Leader approval is recommended for unclear specifications, API impact, security risk, billing impact, or data deletion."
},
{
"question": "Does it include report templates?",
"answer": "Yes. It includes task reporting and child-agent launch log templates."
},
{
"question": "How does it improve safety?",
"answer": "It defines role boundaries, escalation points, review steps, and shutdown expectations for unused child agents."
}
]
},
"file_structure": [
{
"name": "SKILL.md",
"type": "file",
"path": "SKILL.md",
"lines": 80
}
]
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |