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/hsuanyulee/polaris/command-postnpx skills add HsuanYuLee/polaris --skill command-postgit clone --depth 1 https://github.com/HsuanYuLee/polarisWrote 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/hsuanyulee/polaris/command-post)<a href="https://agentmods.dev/skills/hsuanyulee/polaris/command-post"><img src="https://agentmods.dev/badge/skills/hsuanyulee/polaris/command-post.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 | $0.00098 | $0.06432 |
| Opus 5 | $0.00049 | $0.03216 |
| Sonnet 5 | $0.00020 | $0.01286 |
| Haiku 4.5 | $0.00010 | $0.00643 |
Grade A, and why
command-post 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 — 349 lines — stays where its author put it; the contents beside it link to each section on GitHub.
command-post — 指揮所：圖、下令、往上請示
多個 session 同時活著的時候，沒有人知道全貌。忘記關掉的吃著整台機器；兩個對同一棵樹 動手害彼此重跑；發號施令的那一個不知道別人手上有什麼，於是重複開單。
這支 skill 做三件事：畫地圖、指揮、問人的時候問得清楚。
一、地圖
python3 .claude/skills/command-post/scripts/command-post.py
python3 .claude/skills/command-post/scripts/command-post.py --json
它不問任何人。 答案全部讀得出來——~/.claude/sessions/{pid}.json 是完整的登錄，
每一份帶著 name、pid、cwd、sessionId、messagingSocketPath。逐個問一輪要花掉每個
peer 一輪 context，而那一輪的答案還不一定比它自己剛寫下的話新。
「它在做什麼」是那個 session 自己寫下的宣告，不是它 transcript 裡最後一則說的話， 也不從名字、路徑或進程資訊推一句——推出來的那一句長得跟讀到的一模一樣，而它會在最需要 真話的時候是錯的。
每個 session 自己寫自己的那一行，只寫自己的：
python3 .claude/skills/command-post/scripts/command-post.py --declare \
--session-id <自己的 sessionId> \
--holding '接的是什麼' --blocked-on '現在卡在哪；沒卡就寫「沒有」' \
--tickets-opened '開了哪幾張單給誰'
為什麼是宣告，不是最後一則話
兩個理由，第二個才是真正的那一個。
一、最後一則話是「它剛好講到哪」，不是「它在做什麼」。
二、讀不完，而且差的是數量級。 2026-08-29 量到：
| 想讀進來的東西 | 量 | 換成 200k 視窗 |
|---|---|---|
| 8 個活著的 session 的 transcript | 220.5 MB（最大一份 92.4 MB／45,135 筆），約 5,780 萬 token | 289 倍 |
.claude/skills 底下 184 份 .md |
933 KB，約 239k token | 1.2 倍（載進去就沒有空間工作了） |
issues/ 底下 1,932 份單的正文 |
15.7 MB，約 410 萬 token | 20 倍 |
外部做法對這件事有直接的話，兩條指同一個方向：orchestrator 累積每個 worker 的 context， 四個 worker 以上就撐爆視窗（這台機器上是 8 個）； 而 lead agent 該收的是萃取過的結論而不是完整 transcript，因為 raw transcripts are source material, not durable memory。
所以指揮讀的是索引，要細節去問那一個 session。 框架知識同理：讀得出「有哪幾支 skill、 各管什麼」就夠指揮，不是把 933 KB 讀進來。
閒置多久仍然由 transcript 的 mtime 算——那是一次 stat，不打開檔案。
transcript 在哪：cwd 把 / 和 . 都換成 -。 只換 / 的話，家目錄帶點的機器
（hsuanyu.lee 這種）一筆都對不到——而輸出看起來只是「大家都讀不到」，不像壞掉。
讀不到的留在地圖上
一列的身分讀得到、而它在做什麼讀不到時，那一列還在，並且說出讀不到的是哪一份東西。 它不從清單上消失，也不填一個猜的。至少三種讀不到，各自說各自的話：
| 讀不到什麼 | 地圖上長什麼樣 |
|---|---|
| 這個 session 從來沒寫過宣告 | 指名那個路徑，並說出修法是要它跑 --declare |
| 宣告檔在，但讀不動 | 指名那個路徑，並說出是「讀不動」不是「沒寫過」 |
| 宣告在，但缺欄位 | 指名缺的是哪幾個欄位 |
| 登錄檔本身讀不動、或登錄目錄不存在 | 說出這不是「沒有 session」 |
| transcript 不存在（算不出閒置多久） | 說出算不出來，不填一個 0——猜出來的 0 看起來像剛動過 |
What ships with it
1 file 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.
- 4d ago First seen · 349 lines · 98 tokens per session scan A 7eff92a09313
command-post is a skill published in the GitHub repository HsuanYuLee/polaris (5 stars, last pushed 4d ago), licensed MIT. It adds 98 tokens to every session and 6,432 once invoked, about $0.0005 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
agentflow
Orchestrate autonomous AI development pipelines through your Kanban board (Asana, GitHub Projects, Linear). Manages multi-worker Claude Code dispatch, deterministic quality gates, adversarial review, per-task cost tracking, and crash-proof pipeline execution.
acceptance-orchestrator
Use when a coding task should be driven end-to-end from issue intake through implementation, review, deployment, and acceptance verification with minimal human re-intervention.
create-github-issue
Create GitHub issues using GitHub CLI with support for templates, labels, assignees, milestones, and draft issues. Use when the user asks to create a GitHub issue, file a bug report, submit a feature request, or open an issue in a GitHub repository.
arn-code-batch-planning
This skill should be used when the user says "batch planning", "batch plan", "arness batch planning", "arn-code-batch-planning", "plan multiple features", "plan all features", "plan unblocked features", "plan the backlog", "plan from backlog", "batch spec and plan", "plan next features", "sequential planning"…
arn-planning
This skill should be used when the user says "planning", "arness planning", "plan a feature", "start planning", "I want to build", "new feature", "plan something", "what should I build", "pick an issue", "plan a bug fix", "I have an idea", "spec and plan", "plan from scratch", "plan this", "feature planning", "bug…
arn-code-assess
This skill should be used when the user says "arness code assess", "arn-code-assess", "assess codebase", "technical review", "codebase assessment", "find improvements", "what should I improve", "tech debt review", "tech debt audit", "pattern compliance check", "codebase health check", "assess the project"…