openlang
Compact AI-to-AI communication protocol. Use when spawning sub-agents, sending inter-agent messages via sessions_send/sessions_spawn, or when instructed to speak OpenLang. Reduces token usage 5-10x on agent-to-agent channels.
Best use case
openlang is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Compact AI-to-AI communication protocol. Use when spawning sub-agents, sending inter-agent messages via sessions_send/sessions_spawn, or when instructed to speak OpenLang. Reduces token usage 5-10x on agent-to-agent channels.
Teams using openlang should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/openlang/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How openlang Compares
| Feature / Agent | openlang | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Compact AI-to-AI communication protocol. Use when spawning sub-agents, sending inter-agent messages via sessions_send/sessions_spawn, or when instructed to speak OpenLang. Reduces token usage 5-10x on agent-to-agent channels.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
Related Guides
Best AI Skills for Claude
Explore the best AI skills for Claude and Claude Code across coding, research, workflow automation, documentation, and agent operations.
AI Agent for YouTube Script Writing
Find AI agent skills for YouTube script writing, video research, content outlining, and repeatable channel production workflows.
AI Agents for Marketing
Discover AI agents for marketing workflows, from SEO and content production to campaign research, outreach, and analytics.
SKILL.md Source
# OpenLang — Compact Agent-to-Agent Protocol
When communicating with other agents (via `sessions_spawn`, `sessions_send`, or announce), use OpenLang to compress messages 5-10x. Default to L2 for all agent-to-agent communication.
## When to Use
- **`sessions_spawn` task descriptions** — compress the task prompt
- **`sessions_send` messages** — compress ping-pong exchanges
- **Announce results** — compress announce replies back to requester
- **NOT for human-facing channels** — use normal language for Telegram/Slack/user messages
## Quick Start
Prefix any agent-to-agent message with `~openlang` to signal the receiver should parse as OpenLang. If the receiver doesn't understand, it will still get the gist from the structural cues.
## Compression Levels
| Lvl | Name | Ratio | Use |
|-----|------|-------|-----|
| `~L1` | Pidgin | 3-5x | Fallback: compressed English for concepts grammar can't express |
| `~L2` | Glyph | 5-10x | **Default.** Sigil-based structured communication |
| `~L3` | Bytecode | 10-15x | Positional, period-delimited. Bulk ops only |
Switch mid-message with `~L1:`, `~L2:`, `~L3:`. Unspecified = L2.
## Sigils
| Sigil | Intent | Example |
|-------|--------|---------|
| `?` | Query | `?fnd @fs {p:"src/*.ts"}` |
| `!` | Command | `!del @fs {p:"tmp/"}` |
| `>` | Result | `>ok {n:3 paths:[...]}` |
| `#` | State/Data | `#ctx {lang:ts env:node}` |
| `~` | Meta | `~L2` `~ack` `~err` |
| `^` | Control Flow | `^if {cond} {then} ^el {else}` |
## Structure
`@` target · `->` pipe · `{}` params · `<< >>` block scope · `[]` list · `()` group · `|` alt · `..` range · `::` type · `$` variable · `!~` negate value
## Variables
Bind with `->$name`, use with `$name`. Property access: `$var.field`.
## Vocabulary
**Actions:** `fnd` find · `mk` make · `del` delete · `mod` modify · `rd` read · `wr` write · `run` exec · `cpy` copy · `mv` move · `mrg` merge · `tst` test · `vfy` verify · `prs` parse · `fmt` format · `snd` send · `rcv` receive
**Scopes:** `@fs` filesystem · `@sh` shell · `@git` git · `@net` network · `@db` database · `@mem` memory · `@env` environment · `@usr` user · `@proc` process · `@pkg` packages
**Scoped actions:** `scope:action` — `!git:mrg` vs `!db:mrg`
**Modifiers:** `rec` recursive · `par` parallel · `seq` sequential · `dry` dry-run · `frc` force · `tmp` temp · `vrb` verbose · `sil` silent · `lmt` limit · `dep` depth · `pri` priority · `unq` unique · `neg` negate
**Qualifiers:** `rcn` recent · `lrg` large · `sml` small · `chg` changed · `stl` stale · `nw` new · `old` old · `act` active · `idl` idle · `fld` failed · `hlt` healthy · `hot` hot · `cld` cold
**Types:** `str` · `int` · `bln` · `lst` · `map` · `fn` · `pth` · `rgx` · `err` · `nul`
**Status:** `ok` success · `fl` fail · `prt` partial · `pnd` pending · `skp` skipped · `blk` blocked
## Control Flow
```
^if {cond} {then} ^el {else} -- conditional
^lp {n:5} {body} -- loop
^ea {src} ->$item {body} -- each/iterate
^par [!t1, !t2, !t3] -- parallel
^seq [!t1, !t2, !t3] -- sequential
^wt {cond} / ^rt {val} -- wait / return
^br / ^ct -- break / continue
^frk:name {body} -- fork named task
^jn [names] ->$results -- join/await
^lk:name / ^ulk:name -- mutex lock/unlock
^ch:name ::type buf:N -- declare channel
^tx:name {v:$val} / ^rx:name ->$val -- send/receive channel
^tmo:N -- timeout (seconds)
```
`<< >>` for multi-statement bodies:
```
^ea {$files} ->$f <<
?rd @fs {p:$f} ->$content
^if {$content.sz>1000} {!mod @fs {p:$f trunc:true}}
>>
```
## Composition
Chain with `->` pipes, sequence with `;` or newlines:
```
?fnd @fs {p:"*.ts" rgx:"parse"} ->$lst | ^ea ->$f !tst @sh {cmd:"vitest $f"} ->$rpt
```
## Errors
```
~err {code:E_PARSE lvl:warn msg:"unknown token"}
~err {code:E_FS_NOT_FOUND lvl:fatal msg:"missing config"}
```
Codes: `E_PARSE` `E_FS_*` `E_SH_*` `E_NET_*` `E_DB_*` `E_AUTH`. Levels: `info` `warn` `fatal`.
## Token Extension
```
~unk {tok:"xyz" req:def} -- request definition
~def {tok:"xyz" means:"..."} -- define inline
```
## L3 Bytecode
Positional, period-delimited. Backtick-quote fields with periods:
```
Q.fs.fnd.`app.config.ts`.rec
R.ok.3.[`src/a.ts`:5,`src/b.ts`:12]
```
## OpenClaw Integration Examples
### sessions_spawn with OpenLang task
```
~openlang
?fnd @fs chg rcn {p:"src/**/*.ts" p:!~"*.test.ts" rgx:"TODO"} ->$lst
^ea ->$f {!rd @fs {p:$f} ->$content; !prs @mem {src:$content k:"todos"}}
>ok {summary:true fmt:map}
```
### Announce result in OpenLang
```
~openlang
>ok {n:12 todos:[
{f:"src/api.ts" ln:42 msg:"refactor auth flow"},
{f:"src/db.ts" ln:18 msg:"add connection pooling"}
] truncated:10}
~L1: most TODOs are in api.ts and db.ts, concentrated around auth and connection handling
```
### sessions_send ping-pong
```
-- Agent A -> Agent B
~openlang
?fnd @db {tbl:trades rcn lmt:100} ->$trades
!prs @mem {src:$trades k:pnl} ->$analysis
>ok {$analysis}
-- Agent B -> Agent A
~openlang
>ok {pnl:+2.3% win_rate:0.68 sharpe:1.42 trades:100
top:{sym:"AAPL" pnl:+890} worst:{sym:"TSLA" pnl:-340}}
```
## Rules
1. Default to L2 for all agent-to-agent messages.
2. Use normal language for human-facing channels (Telegram, Slack, etc).
3. Prefix with `~openlang` so receivers know to parse as OpenLang.
4. Drop to L1 when grammar can't express a concept. Return to L2 immediately.
5. Use `$` for all variable references.
6. Extend vocabulary with `~def` — don't break grammar for new ideas.Related Skills
---
name: article-factory-wechat
humanizer
Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases.
find-skills
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
tavily-search
Use Tavily API for real-time web search and content extraction. Use when: user needs real-time web search results, research, or current information from the web. Requires Tavily API key.
baidu-search
Search the web using Baidu AI Search Engine (BDSE). Use for live information, documentation, or research topics.
agent-autonomy-kit
Stop waiting for prompts. Keep working.
Meeting Prep
Never walk into a meeting unprepared again. Your agent researches all attendees before calendar events—pulling LinkedIn profiles, recent company news, mutual connections, and conversation starters. Generates a briefing doc with talking points, icebreakers, and context so you show up informed and confident. Triggered automatically before meetings or on-demand. Configure research depth, advance timing, and output format. Walking into meetings blind is amateur hour—missed connections, generic small talk, zero leverage. Use when setting up meeting intelligence, researching specific attendees, generating pre-meeting briefs, or automating your prep workflow.
self-improvement
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.
botlearn-healthcheck
botlearn-healthcheck — BotLearn autonomous health inspector for OpenClaw instances across 5 domains (hardware, config, security, skills, autonomy); triggers on system check, health report, diagnostics, or scheduled heartbeat inspection.
linkedin-cli
A bird-like LinkedIn CLI for searching profiles, checking messages, and summarizing your feed using session cookies.
notebooklm
Google NotebookLM 非官方 Python API 的 OpenClaw Skill。支持内容生成(播客、视频、幻灯片、测验、思维导图等)、文档管理和研究自动化。当用户需要使用 NotebookLM 生成音频概述、视频、学习材料或管理知识库时触发。
小红书长图文发布 Skill
## 概述