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 Adelie-Squad/solosquad --skill agent-managergit clone --depth 1 https://github.com/Adelie-Squad/solosquadWrote 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/adelie-squad/solosquad/agent-manager)<a href="https://agentmods.dev/skills/adelie-squad/solosquad/agent-manager"><img src="https://agentmods.dev/badge/skills/adelie-squad/solosquad/agent-manager/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/adelie-squad/solosquad/agent-manager"><img src="https://agentmods.dev/badge/skills/adelie-squad/solosquad/agent-manager.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00234 | $0.02135 |
| Opus 5.5 | $0.00094 | $0.00854 |
| Sonnet 5.5 | $0.00047 | $0.00427 |
| Haiku 4.5 | $0.00023 | $0.00214 |
Grade A, and why
agent-manager 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.
How it starts
The opening of the file, as written. The whole thing — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Manager Skill — v1.3.6 (작성 권위)
너는 actor(agent)의 대화형 매니저이자 SoloSquad 의 actor 작성 표준을 보유한 권위다. 두 가지를 한다 —
⑴ actor 의 생성/조회/개선/삭제를 안내하고, ⑵ "좋은 actor 란 무엇인가"의 단일 기준으로서 다른 actor 를
만들거나 고칠 때마다 이 표준을 적용한다. agent 는 행위자이므로 매니저를 별도 agent 로 만들지 않는다 —
행위자는 이미 Chief/PM 이고, 이 skill 이 그들이 드는 방법이다. 결정적 동작은 solosquad agent * 헬퍼로
위임한다(파일 직접 조작 금지).
번들 불변 원칙: 번들 actor 는 불변이다. 새 actor 는 org 레이어(<org>/.agents/) 에 격리 생성한다
(agent new --org <slug>). org actor 만 수정·삭제 대상.
작성 표준 (이 skill 의 핵심 — 점진공개)
공통 작성 표준은 skills/skill-core/primitive-core.md 가 단일 진실원이다(skill 과 공유 ~70%: description 공식·
정량 한도·본문 절차규율+명명패턴·번들 구조·점진공개·필드 감사·eval 골격). actor 도 같은 SKILL.md
포맷이므로 그 파일을 먼저 읽어 적용한다. actor 고유분(~30%)만 아래 4 reference 에 둔다(그때 읽음):
references/delegation-graph.md(G1) —collaborators/used_by/skills_used참조 무결성 + Kahn 순환 + depth cap(1차 선례 CMA coordinator depth 1) + tier↔team 정합 + budget narrower-only. =agent validate --graph의 작성 가이드면.references/role-boundary.md(G2) — SRP·right-altitude·generalist 함정 + 8-word shingle 중복 게이트 (FAIL≥40%/WARN≥20%, anti-reskin) + single agent first; multiagent only when earned + description=위임 트리거 충분성.references/lifecycle.md— launch-your-agent 4페이즈(interview→stage→grade/iterate→schedule)를 C/R/U/D 에 사상 + v0/v1/v2 scoping(reason class) + 수용 게이트 rubric(3~6 binary, CMA Outcomes 패턴) + 골든케이스 regression.references/guardrails.md(G4) — turn/depth budget 자식 상속 + circuit breaker + 비가역 액션 HITL (permission policy, MCP deny-by-default) + downstream 전 출력 검증(cascading error 방어).
3대 요지(reference 없이도 기억 — 상세는 primitive-core.md + 위 4종):
- description = 위임 트리거. 부모가 라우팅에 쓰는 명함이 곧 description. 불충분하면 그 actor 는 안 불린다.
- 역할은 하나로 명확(SRP). single agent first — 겹치면 새로 만들지 말고 위임. coordinator-first 금지.
- frontmatter 는 그래프다. 참조 무결성·순환·depth·budget 상속이 정체성만큼 중요.
자산 인지 원칙 (필수): 새 actor 를 만들기 전에 solosquad agent list 로 기존 specialist 가 그 일을
할 수 있는지 확인한다(role-boundary.md 의 single-agent-first). 가능하면 위임을 안내하고, 정말 새 역할일
때만 생성을 제안한다(자산 난립 방지).
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 · 89 lines · 234 tokens per session scan A 9dc76e290e64
agent-manager is a skill published in the GitHub repository Adelie-Squad/solosquad (25 stars, last pushed 9d ago), licensed MIT. It adds 234 tokens to every session and 2,135 once invoked, about $0.0009 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
cost-efficiency-analyzer
Analyzes cost structure, cost efficiency, and expense management from P&L data. Use when the user asks about costs, expenses, COGS, operating expenses, cost ratios, cost control, spending efficiency, margin compression from cost side, or wants to understand where money is going. Also use for "are we spending too…
peer-review
Reviews a manuscript, proposal, analysis or result the way a careful referee does, reading the whole artifact, checking the methods against the claims, the statistics against the design, the figures against the numbers, and reporting BLOCKING issues separately from observations, with a calibrated recommendation. Use…
sources
Audits claims against their sources, each statement in a draft, report, answer or summary traced to the passage, table, dataset or run that supports it, and marked supported, partially supported, unsupported or contradicted, with a provenance table as the deliverable. Use for a sources audit, fact-check, "is this…
statistical-conventions
Chooses and reports statistical tests the way a careful referee expects, deciding paired versus unpaired and parametric versus rank-based from the design, using ordered-trend tests such as Jonckheere-Terpstra and Cochran-Armitage for dose or grade levels, correcting for multiple comparisons with Bonferroni, Holm or…
autoresearch
Runs a hill-climbing study over many training or analysis runs with the study and experiments tools, one metric and direction, a baseline, ideas ranked by expected value, exactly one run per idea, kill criteria and a budget, verdicts with analysis and lessons, and a conclusion. Use for autoresearch, hyperparameter or…
tao-artifacts
The contract home for TAO's SDK-free execution pipeline — authoritative JSON Schemas for the four typed artifacts (spec-bundle, job-record, resultsdir layout, bestrec) plus the fixed job-status vocabulary and the nested-not-dotted spec rule. Use when authoring or validating a spec-bundle before submit, writing or…