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 0fuk/OpenACCP --skill compact-sgit clone --depth 1 https://github.com/0fuk/OpenACCPWrote 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/0fuk/openaccp/compact-s)<a href="https://agentmods.dev/skills/0fuk/openaccp/compact-s"><img src="https://agentmods.dev/badge/skills/0fuk/openaccp/compact-s/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/0fuk/openaccp/compact-s"><img src="https://agentmods.dev/badge/skills/0fuk/openaccp/compact-s.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 | $0.00120 | $0.02210 |
| Opus 5.5 | $0.00048 | $0.00884 |
| Sonnet 5.5 | $0.00024 | $0.00442 |
| Haiku 4.5 | $0.00012 | $0.00221 |
Grade A, and why
compact-s 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 2d 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 — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/compact-s — 先外置,再压缩:compact 不再丢东西
平台自动 compact 是盲目摘要——owner 原话被压成「做了一些决定」。本 skill 不挑刺复核 compact,而是让它无所谓:凡只活在对话里的高价值状态,先高保真落盘。项目事实(协调目录/台账路径)读项目根 PROJECT-PROFILE.md。
0 · 判据:什么进 checkpoint
唯一标准:这条信息是否只存在于本会话对话流里? 已落台账/memory/协调目录的只登记路径+一句钩子;本会话新发生未外置的逐字转录;判断不了当作只在对话里。
1 · Checkpoint(固定结构,高保真区禁摘要)
路径:值守项目=<coordination>/<角色>-COMPACT-CHECKPOINT-<YYYYMMDD-HHMM>.md;非项目会话=args 指定目录或 ~/.claude/compact-checkpoints/<cwd-slug>-<时间戳>.md。
- owner 决策:本会话每次拍板/裁决/授权的原话+时刻;已入台账的只记条号+路径。
- 编排计划与目标态:值守计划/交接件的目标态原文+路径;基面 SHA/tag/冻结态等会话内刚更新的数字;走五门发版范式的项目必写当前门次:本车在第几门、哪些卡缺哪一门、五门台账(工单集)绝对路径——compact 后第一动作就靠这行定位。
- 验收目标与宪法级判准:硬门清单+「合流≠验收/回归门≠决策完成度」类判准原文+出处。
- 关键文件路径:本会话读过/写过的载重文件绝对路径,一行一个+一句说明。
- 在飞与叫醒:在飞表(开放目标 + 执行体:子代理写 agentId 与 output 文件路径,跨会话写会话名与候回执件名 + 最近动作 mtime)+ 已挂 ScheduleWakeup 的时刻与 prompt 原文(没挂 = 违规,先挂再压)+ 本窗跑偏 / 违规记录。叫醒机制只认两种:子代理完成通知 / 跨会话直推(主信号)与 ScheduleWakeup(兜底);Monitor 文件信箱仅在项目仍用它时登记。
可压缩区:一段话列出可被摘要压掉的工程往返(已固化的命令输出/报错现场)。 纪律:零编造;记不清原话就 Read 回原件;写不满如实写「无此类增量」。
2 · 压缩(checkpoint 落盘并亲验之后)
- ls/Read 回首尾亲验。
- /compact 是否吃自定义指引因版本而异,现场验证;不确定就普通 compact,checkpoint 在盲压也不丢。
- 模型不能自己执行 /compact;自动 compact 逼近时不等 owner,立刻落 checkpoint。
2.5 · 固定交付:两段可粘贴 prompt(回复末尾原样给出,尖括号换真值,不加解释;20261010 修订)
A. compact 指引(给平台的压缩指令:只说留什么、压什么,一行)
/compact 保留:owner 决策原话与时刻;当前门次 + 基面 SHA/tag;在飞子代理 agentId 与其 output 文件路径、在飞跨会话令与候回执件名;已挂 ScheduleWakeup 时刻;验收硬门原文;关键文件绝对路径。可压:已固化进文件的命令输出 / 报错现场 / 被拦细节 / 已收货的交回正文。checkpoint=<checkpoint 绝对路径>
B. compact 后启动短 prompt(给自己的第一条指令:先定位、再看活性、再挂表、再干活;不重做一次性 setup)
你是 <角色>。按序做,不问我:① date;全文读 <checkpoint 绝对路径>,摘要与它冲突以它为准。② 按 checkpoint 时刻前 30 分钟至今的 mtime 窗口 ls 文件信箱,逐件补收。③ 列在飞执行体:每个子代理 ls 其 output 文件 mtime、每条跨会话令 ls 其回执 / 进行中件 mtime;> 30 分钟无变化的按 /keep-pushing §7 停掉重派;然后挂 ≤ 30 分钟 ScheduleWakeup。④ 读五门台账定位门次。⑤ 向我发一条「记忆恢复确认」(人话,≤ 8 行):当前目标链(到哪一门、下一步是什么、终点是什么)、在飞几人各几分钟没动、候我的事(没有就写没有)、第一批动作头三条;与 checkpoint 对不上的地方如实标出。⑥ 不等我回复,按 checkpoint「compact 后第一批动作」逐条执行。不重跑已做过的一次性动作(建 Monitor、写值守计划、通告 lane)。
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.
- 2d ago First seen · 55 lines · 120 tokens per session scan A f5d9d977d09a
compact-s is a skill published in the GitHub repository 0fuk/OpenACCP (2 stars, last pushed 2d ago), licensed MIT. It adds 120 tokens to every session and 2,210 once invoked, about $0.0005 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 2026-10-14.
Other skills, from other repositories
hive-mind-advanced
Queen-led collective-intelligence coordination with consensus mechanisms and persistent shared memory. Use when orchestrating a large swarm of agents under a lead coordinator, building consensus across agents, or maintaining durable cross-agent memory.
sema-usage
Content-addressed vocabulary protocol via the sema MCP server. Use when reasoning through a problem that a named pattern could structure, such as a decision, design, diagnosis, forecast or disagreement, and when working with shared vocabulary: searching, resolving, minting, and verifying meaning across agents and…
hive-mind-advanced
Advanced Hive Mind collective intelligence system for queen-led multi-agent coordination with consensus mechanisms and persistent memory.
cross-agent-handoff
Transfer context between AI agent sessions with structured handoff protocols, state serialization, and decision log preservation. Covers multi-agent coordination, context compression, and continuity patterns. Triggers on agent handoff, session transfer, or multi-agent continuity requests.
hive-mind-advanced
Advanced Hive Mind collective intelligence system for queen-led multi-agent coordination with consensus mechanisms and persistent memory.
handoff-mesh
Built-in communication layer for multi-agent setups. Full-mesh handoff folders (any agent can hand off to any other), an hourly watcher that surfaces open handoffs, weekly archive cleanup, and a completion-reporting loop so every handoff is handled AND verified. The durable 'brain' counterpart to real-time channels…