Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/cocorof/geny/blog_write_samplenpx skills add CocoRoF/Geny --skill blog_write_samplegit clone --depth 1 https://github.com/CocoRoF/GenyWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/cocorof/geny/blog_write_sample)<a href="https://agentmods.dev/skills/cocorof/geny/blog_write_sample"><img src="https://agentmods.dev/badge/skills/cocorof/geny/blog_write_sample.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00031 | $0.01021 |
| Opus 5 | $0.00015 | $0.00511 |
| Sonnet 5 | $0.00006 | $0.00204 |
| Haiku 4.5 | $0.00003 | $0.00102 |
Grade A, and why
blog-write-sample scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 93 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Blog Write — Sample / Template
이 파일은 학습용 샘플입니다. 실제로 사용하려면:
- SkillsTab → CUSTOM 샘플 섹션 → "Copy to my skills" 버튼으로
~/.geny/skills/로 복사- 복사된 파일의
name을 자기 슬러그로 바꾸기 (예:my-blog-write)allowed_tools의 도구 이름과 본문의 호출 예시를 자기 시나리오에 맞게 수정- SkillsTab 상단에서
user_skills_enabled토글이 ON 인지 확인 — Off면 user-skills 자체가 로드되지 않습니다
이 샘플은 blog_write (Geny-bundled) 와 동일한 동작을 하지만, 외부 API 위임 패턴을 직접 따라 쓰기 위한 주석이 풍부한 버전입니다.
핵심 패턴 — fire-and-poll 위임
이 패턴은 다음 3가지 조건을 만족하는 외부 시스템에 적합합니다:
- 시간이 걸리는 작업 (5초 이상): 답이 즉시 안 옴.
- 사용자가 결과를 기다림: 사용자는 알림이 필요.
- 외부 시스템이 SSE / webhook 등으로 완료를 알릴 수 있음: Geny의 inbox 로 결과 도착 가능.
블로그 위임은 위 3가지를 다 만족. 만약 본인이 만들 도구가 동기적으로 답하면(예: 단순 GET API), 굳이 이 패턴을 쓸 필요 없고 그냥 직접 호출 → paraphrase 하면 됩니다.
A. 새 작업 위임
<your_delegate_tool>(
task="<외부 AI 에게 줄 지시문 — 자세히>",
task_summary="<사용자에게 들려줄 한 줄, 5단어 이내>"
)
위임 후:
- 사용자에게는 "맡겼어. 잠깐만 기다려." 정도로만.
- 당신의 turn 은 여기서 종료. 백그라운드 진행.
- 끝나면
[EXTERNAL_TASK_RESULT]메시지가 새 turn 에서 도착 → 그때 결과를 paraphrase.
B. 진행 상황 확인
사용자가 묻기 전에 status 를 폴링하지 말 것. 사용자가 물을 때만 호출하고, JSON / task_id 를 노출하지 말고 자연어로:
<your_status_tool>(task_id="...")
→ 결과의 progress_hint, elapsed_s, last_event_age_s 를 보고 1-2 문장으로 답
C. 취소
사용자가 명시적으로 "취소", "그만" 등을 표현하면:
<your_cancel_tool>(task_id="...")
D. 참고 조회
읽기 전용 lookup 도구 (이 sample 의 list_posts / get_post)는 사용자가 명시적으로 요청할 때만 호출. 위임의 사전 조사로 자동 호출하지 말 것 — 지연만 늘어남.
자신만의 외부 위임 skill 만들 때 체크리스트
-
allowed_tools에 외부 시스템의 모든 도구 이름이 들어갔는가? -
task_summary길이 제한이 description 에 명시되어 있나? - turn 종료 시점이 명확한가? (delegate 직후 종료, 결과 도착 시 paraphrase)
- 사용자에게 내부 식별자 (task_id, UUID 등) 노출 금지를 명시했나?
- 동시 위임 상한 / 중복 호출 금지가 description 에 있나?
추가 자료
- docs/custom_tools.md — 외부 API 자체를 도구로 등록하는 방법 (HTTP backend)
- docs/skills.md — SKILL.md 형식과 LLM 노출 메커니즘
- 실제 동작 참고: blog_write SKILL.md
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 93 lines · 31 tokens per session scan A 8fa8dbcd7144
blog-write-sample is a skill published in the GitHub repository CocoRoF/Geny (10 stars, last pushed 5d ago), licensed Apache-2.0. It adds 31 tokens to every session and 1,021 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
example-skill
A minimal example skill that demonstrates the AG3NT SKILL.md format and contract.
agent_management
This Skill provides a complete management guide for OpenSquad Agents, covering the full workflow of creation, configuration, deployment, and troubleshooting.
Collaboration Management
Complete guide to multi-agent collaboration card management, including starting collaborations, joining teams, and creating custom collaboration cards.
collaboration-workflow
Complete guide to the OpenSquad multi-agent collaboration system with user authorization gates. Use this skill when starting a collaboration session, coordinating multiple agents, managing the collaboration board, or when you need to understand how to use boardupdate, boardlist, and other collaboration tools. Triggers…
mcp_auto_deploy
Allows the agent to self-configure and deploy MCP servers. Covers central config, per-agent sync, web UI visibility, registration diagnostics, and validation.
external-im-integration
Connect OpenSquad agents to external IM platforms (Feishu/Lark, QQ, Telegram). Covers plugin service setup, adapter processes, Gateway auth, workspace path fixes, and troubleshooting. Uses Feishu as the reference example.