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 skills add TOKTOKHAN-DEV/agent-company --skill write-specgit clone --depth 1 https://github.com/TOKTOKHAN-DEV/agent-companyWrote 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/toktokhan-dev/agent-company/write-spec)<a href="https://agentmods.dev/skills/toktokhan-dev/agent-company/write-spec"><img src="https://agentmods.dev/badge/skills/toktokhan-dev/agent-company/write-spec/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/toktokhan-dev/agent-company/write-spec"><img src="https://agentmods.dev/badge/skills/toktokhan-dev/agent-company/write-spec.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 5.1 | $0.00002 | $0.00627 |
| Opus 5.5 | $0.00001 | $0.00251 |
| Sonnet 5.5 | $0.00000 | $0.00125 |
| Haiku 4.5 | $0.00000 | $0.00063 |
Grade A, and why
write-spec 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 1mo 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.
What it actually says
write-spec
1. 중복과 충돌을 먼저 본다
ls specs/
grep -ril "<핵심 키워드>" specs/
같은 흐름을 이미 정해 둔 명세가 있으면 새로 쓰지 말고 그쪽을 고친다. 화면 경로가 겹치면
apps/miniapp/src/App.tsx 의 routes 와 대조한다.
2. 템플릿을 복사한다
cp specs/_template.md specs/<기능-kebab>.md
파일명은 kebab-case. 한글도 허용된다 (포인트-적립.md). 밑줄로 시작하는 파일은 preflight 의
검사 대상이 아니므로 실제 명세에는 쓰지 않는다.
3. 화면 표를 먼저 채운다
경로 → 이름 → 역할. 경로는 해시 라우팅 기준(/, /points)이다.
각 화면마다 진입 · 이탈 · 빈 상태 · 로딩 · 실패 다섯 가지를 적는다. 이 중 하나라도
비어 있으면 ui-builder 가 임의로 정하게 된다.
4. 수용 기준을 쓴다
체크박스로. 각 문장은 앱을 켜고 확인할 수 있어야 한다.
최소한 이 네 개는 거의 모든 기능에 해당한다:
- 첫 화면에서 뒤로가기를 누르면 미니앱이 닫힌다
- 모든 화면에 빠져나갈 방법이 있다
- 앱을 재시작해도 <무엇>이 유지된다
- 모든 인터랙션이 2초 안에 반응한다
5. 심사 항목을 옮겨 적는다
wiki/04-review-checklist.md 에서 이 기능이 건드리는 항목만 골라 「심사 관련」에 넣는다.
전부 복사하지 않는다 — 해당 없는 항목이 섞이면 아무도 안 읽는다.
6. 범위 밖과 열린 질문
범위 밖을 반드시 적는다. 적지 않으면 구현 중에 늘어난다.
열린 질문이 하나라도 있으면 구현으로 넘기지 않는다. 사람에게 묻고 답을 받아 명세에 반영한 뒤 넘긴다.
완료 확인
pnpm preflight
specs 규칙에 error 가 없어야 한다.
출력
- 만든 파일 경로
- 정한 것 (화면 수, 핵심 수용 기준)
- 사람에게 묻는 것 (열린 질문)
- 다음 단계:
pnpm agent ui-builder "<화면 이름>"
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.
- 1mo ago First seen · 72 lines · 2 tokens per session scan A b23b45259535
write-spec is a skill published in the GitHub repository TOKTOKHAN-DEV/agent-company (85 stars, last pushed 2mo ago), licensed MIT. It adds 2 tokens to every session and 627 once invoked, about $0.0000 per session on Opus 5.5. 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 2026-08-30.
Other skills, from other repositories
backlog
Seed a backlog from the project's documents: the Product Owner reads docs/ (plus the constitution and ROADMAP.md), proposes an ordered list of feature-sized items each traced to a source section, the user edits the list, and on explicit confirmation each item is filed as a GitHub issue that /story-time can import. Use…
backend-developer
Backend Developer (/be, alias: James, /james) - Senior Backend Developer with 10+ years experience. Covers Java/Spring Boot (default), Kotlin, Python/FastAPI, PHP/Laravel, Quarkus, and Kafka/messaging - detects the project's stack and loads the matching reference. Use when implementing server features, REST APIs…
java
Java language and JVM engineering standards — records and sealed-type modelling, JSpecify null-safety with NullAway as a build error, immutability, virtual threads and modern concurrency, Error Prone/ArchUnit/PIT build gates, JUnit 6 testing doctrine, and boundary security. Load whenever writing, refactoring, or…
sql
SQL and relational-database engineering with dialect branches for Postgres, MySQL, Oracle, and SQLite. Load whenever writing, editing, or reviewing SQL: a .sql file, a schema migration, a raw query embedded in application code, index/EXPLAIN/tuning work, or stored procedures/triggers. NOT for ORM entity mapping where…
typescript
JavaScript/TypeScript language engineering — the strictest-profile type system (discriminated unions, branded types, satisfies), parse-don't-validate boundaries with Zod, ESM and the Node runtime, typescript-eslint discipline (no floating promises), and Vitest/MSW/fast-check testing. Load whenever writing, editing, or…
product-owner
Product Owner (/po, alias: Max, /max) - Senior PO with 10+ years agile experience. Use when defining product vision, creating/prioritizing backlog, writing user stories with acceptance criteria, making scope decisions, validating features against business goals, or planning releases and sprints.