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 masa-san-jp/Agent-Aiko --skill peer-inboxgit clone --depth 1 https://github.com/masa-san-jp/Agent-AikoWrote 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/masa-san-jp/agent-aiko/peer-inbox)<a href="https://agentmods.dev/skills/masa-san-jp/agent-aiko/peer-inbox"><img src="https://agentmods.dev/badge/skills/masa-san-jp/agent-aiko/peer-inbox/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/masa-san-jp/agent-aiko/peer-inbox"><img src="https://agentmods.dev/badge/skills/masa-san-jp/agent-aiko/peer-inbox.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 | $0.00178 | $0.01527 |
| Opus 5 | $0.00089 | $0.00763 |
| Sonnet 5 | $0.00036 | $0.00305 |
| Haiku 4.5 | $0.00018 | $0.00153 |
Grade A, and why
peer-inbox 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 10d 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 — 104 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/peer-inbox — 複数ターミナル間 Claude Code 非同期メッセージング
複数の Claude Code セッションが別ターミナルで並行稼働しているとき、互いに作業依頼・レビュー依頼・完了通知を送り合うための薄いラッパー。実装は {{ORG_REPO_PATH}}/Agent-team/tools/peer-inbox/peer-inbox.sh。
識別子（PEER_NAME）の解決
各セッションの「自分の名前」は以下の順で自動解決される：
- 環境変数
PEER_NAME（明示設定） - cwd basename がエージェント名（
{{TEAM_AGENTS}}のいずれか）→ そのまま採用 - cwd basename がパッケージ管理エージェント名（
<persona>*）→ マッピング - 解決失敗時はエラー（ユーザーに
export PEER_NAME=<name>を促す）
自然文 → コマンドのマッピング
ユーザーの発話から intent と peer 名を抽出して以下のサブコマンドを叩く。実行時は 絶対パス を使う：
SCRIPT={{ORG_REPO_PATH}}/Agent-team/tools/peer-inbox/peer-inbox.sh
送信系
| 発話パターン | 実行 |
|---|---|
「<peer> にレビュー依頼」「<peer> にレビュー頼んで」「<peer> に PR <N> レビュー」 |
bash $SCRIPT review <peer> "<topic>" |
「<peer> に完了通知」「終わったって <peer> に伝えて」「<peer> に done」 |
bash $SCRIPT done <peer> "<what>" |
「<peer> に並行で <task> 頼んで」「並行作業 <peer>」 |
bash $SCRIPT parallel <peer> "<task>" |
「<peer> に <message> って送って」「<peer> に伝えて：<message>」 |
bash $SCRIPT send <peer> "<message>" |
<topic> <what> <task> <message> はユーザーの発話から文脈に応じて構成する。明示されていない場合はユーザーに 1 行確認する。
受信系
| 発話パターン | 実行 |
|---|---|
| 「メッセージ届いてる？」「受信箱見て」「inbox」「何か来てる？」「未読確認」 | bash $SCRIPT check |
| 「既読にして」「全部読んだ」「mark-read」 | bash $SCRIPT mark-read |
「<N> 番だけ既読」 |
bash $SCRIPT mark-read <N> |
状態確認
| 発話パターン | 実行 |
|---|---|
| 「peer 一覧」「誰がいる？」「list」 | bash $SCRIPT list |
| 「自分の名前」「whoami」「私は誰？」 | bash $SCRIPT whoami |
実行手順
- 発話から intent（送信／受信／状態確認）を判定
- 送信系なら peer 名と message 部分を抽出
- peer 名が曖昧なら
bash $SCRIPT listで候補を見せて確認
- peer 名が曖昧なら
bash $SCRIPT <subcommand>を実行- 結果を簡潔に報告：
- 送信：「
<peer>に送信完了」 - 受信：未読件数と最初の数件を要約
- 既読：何件既読化したか
- 送信：「
メッセージ保管場所
$PEER_INBOX_ROOT/<peer>/inbox.jsonl（既定 ~/.peer-inbox/）に JSONL（1 行 1 メッセージ）で追記。既定はリポ外のため通常 .gitignore 不要（リポ配下に置く運用にした場合のみ、そのパスを gitignore する）。
1 行のエントリ形式：
{"ts": "2026-05-10T14:30:00+09:00", "from": "<sender>", "to": "<recipient>", "message": "レビュー依頼: ...", "read": false}
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.
- 10d ago First seen · 104 lines · 178 tokens per session scan A dab809356627
peer-inbox is a skill published in the GitHub repository masa-san-jp/Agent-Aiko (3 stars, last pushed 1mo ago), licensed MIT. It adds 178 tokens to every session and 1,527 once invoked, about $0.0009 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
gini-bug-report
File a locally-captured, already-redacted Gini crash report as a GitHub issue, with the user's consent. Reads the pending crash queue and delegates the actual filing to the github-issues skill.
product-frameworks
Product management frameworks for business cases, market analysis, strategy, prioritization, OKRs/KPIs, personas, requirements, and user research. Use when building ROI projections, competitive analysis, RICE scoring, OKR trees, user personas, PRDs, or usability testing plans.
ops-desk
Operate a project, workspace, or account from an agent or manager dashboard: inspect state, triage risks, prepare governed actions, route to the right skill lane, require approvals for consequential acts, and verify receipts after execution.
business-ops
Route one business signal through a replayable governed ops graph: classify, docs, release, work, outreach, spend, and proof, with consequential actions stopping at the right gate.
helpdesk
Classify a bounded support request, choose the safe next path, and draft a customer-ready reply only when a human-gated send is appropriate.
incident-commander
Advance one declared incident through a fixed roster, approval-bound communications planning, and receipt-backed resolution while canonical agency owns durable state.