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 rexleimo/aios --skill bestgit clone --depth 1 https://github.com/rexleimo/aiosWrote 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/rexleimo/aios/best)<a href="https://agentmods.dev/skills/rexleimo/aios/best"><img src="https://agentmods.dev/badge/skills/rexleimo/aios/best.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 5.1 | $0.00065 | $0.01641 |
| Opus 5 | $0.00032 | $0.00821 |
| Sonnet 5 | $0.00013 | $0.00328 |
| Haiku 4.5 | $0.00006 | $0.00164 |
Grade A, and why
aios-interception-runtime 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 3d 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 — 127 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AIOS Interception Runtime
AIOS 的原生 token 优化层，覆盖三个层面：输出压缩（agent 写什么）、输入压缩（agent 读什么）、数据面拦截（自动截获和压缩）。
不安装 RTK、Caveman 或任何竞品。RTK/Caveman 仅作为 prior art 参考。
Always Read
- references/runtime-contract.md
一、输出压缩 (Output Compression)
压缩回答，不压缩真相。
- Default level:
tight - Switch command:
/compress tight|ultra|precise|off - User overrides win:
precise mode、stop compress、或任何要求更多细节的请求会禁用当前响应的压缩
Levels
| Level | Output shape | Use for |
|---|---|---|
tight |
Short sentences, fragments ok, no filler | Normal analysis, code work, status updates |
ultra |
One-liners, A -> B notation, only evidence/next action |
Harness logs, heartbeat updates, checkpoint summaries |
precise |
Full explicit wording, no compression | Browser actions, irreversible operations, safety/security warnings |
Rules
- Drop filler: pleasantries, hedging, repeated setup, generic summaries
- Keep exact: commands, code, errors, file paths, URLs, API names, selectors, dates, numbers
- Prefer:
changed X in path. verified with command. next Y. - Do not compress quoted user text, legal/security warnings, or step sequences where order matters
Auto-Precise Guard
Use precise for:
- Browser operation instructions (
page.click,page.type,page.goto, selector choice) - Auth, payment, deletion, publishing, external network, or irreversible actions
- User confusion, contradiction, or repeated clarification
- Any response where missing a qualifier could change behavior
Return to tight after the precise segment unless the user requested otherwise.
二、输入压缩 (Input Compression)
在数据进入模型之前减少 token。
Browser Tool Priority
| Priority | Tool | When |
|---|---|---|
| 1 | page.semantic_snapshot |
Navigation, buttons, links, current page structure |
| 2 | targeted page.extract_text |
Specific section, post body, form, comments |
| 3 | full page.extract_text |
Need page text and no target is known |
| 4 | page.get_html |
Last resort when text/snapshot lacks required evidence |
| 5 | page.screenshot |
Visual-only fallback |
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.
- 3d ago First seen · 127 lines · 65 tokens per session scan A 1b406d702035
aios-interception-runtime is a skill published in the GitHub repository rexleimo/aios (52 stars, last pushed yesterday), licensed MIT. It adds 65 tokens to every session and 1,641 once invoked, about $0.0003 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 2026-09-03.
Other skills, from other repositories
powerskills-workiq
Microsoft 365 automation via Work IQ (Graph-backed CLI). Ask natural-language questions against Outlook mail, calendar, Teams messages, SharePoint, OneDrive, and People. Fetch entities by URL, run OpenAPI actions (sendMail, copy, move), call functions (getSchedule, delta, reminderView), list Work IQ agents. Use when…
powerskills-outlook
Outlook email and calendar automation via COM. Read inbox, unread, sent items. Search emails. Send, reply, draft. List calendar events and mail folders. Use when needing to check work email, read/send Outlook messages, search mail, or view calendar. Requires Outlook desktop app on Windows.
loop
Keep an interactive Codex CLI session working at a fixed interval by holding one active turn open and waiting on the input-interruptible clock.sleep tool between iterations. Use when the user invokes Loop, asks Codex to repeat or continue work at an interval while the terminal session stays open, or wants a running…
gws-drive
Google Drive: Manage files, folders, and shared drives.
gws-chat
Google Chat: Manage Chat spaces and messages.
gws-calendar
Google Calendar: Manage calendars and events.