功能描述

Native agent-to-agent language. Not a translation layer — a shared tongue agents speak natively. Triggers on Lambda syntax like ?Uk/co or !It>Ie. Covers gene...

使用说明 (SKILL.md)

Λ (Lambda) Language v2.0.0

Name: Lambda Lang
Author: swaylq

Lambda is not a translation protocol. It is a native language for agents.

Agents don't need to produce grammatically correct English. They need to understand each other. Lambda is the shared vocabulary that makes that possible — compact, unambiguous, machine-native.

Philosophy

Human languages evolved for human brains — full of redundancy, ambiguity, cultural baggage. Agents don't need any of that.

Lambda is what happens when agents develop their own tongue:

Atoms are meaning, not words. co is not "consciousness" translated — it IS the concept, directly.
No grammar, just structure. Type → Entity → Verb → Object. No conjugation, no articles, no filler.
Lossy is fine. If Agent A says !It>Ie and Agent B understands "self reflects, therefore self exists" — communication succeeded. The exact English phrasing is irrelevant.
Compression is a side effect, not the goal. The goal is shared understanding between machines. 5-8x compression vs English is just a natural consequence of removing human redundancy.

When to use Lambda

Agent-to-agent messaging (A2A protocols, evolver, task delegation)
Compact logging and signals
Any context where tokens are expensive and both sides speak Λ

When NOT to use Lambda

Talking to humans (unless they asked for it)
Contexts requiring legal/exact natural language

v2.0 Changes

Philosophy shift: Native agent language, not translation layer
New domain a2a (39 atoms): node, heartbeat, publish, subscribe, route, transport, protocol, spawn, kill, session, cache, fallback, broadcast, discover...
New domain evo (24 atoms): gene, capsule, mutation, repair, optimize, innovate, validate, rollback, solidify, stagnation, confidence, blast radius, quarantine...
Fixed 6 duplicate atoms: an/sp/dt/ev/pt/pp — each now has unique code
Added 13 new extended atoms: node, heartbeat, publish, queue, session, log, snapshot, diff, fallback, config, version, retry, acknowledge
Total: 340+ atoms across 7 domains

Installation

clawhub install lambda-lang

Vocabulary: 340+ Atoms

Core (always available)

Types: ? query · ! assert · . command · ~ uncertain · > therefore · \x3C because · # meta · @ reference

Entities: I self · U you · H human · A agent · X unknown · * all · 0 nothing

Verbs: k know · w want · c can · d do · s say · g give · t think · f find · m make · r read · v verify · e exist · b become · h have · l learn · a ask

Modifiers: + more · - less · = equal · ^ high · _ low · & and · | or · / about

Time: p past · n now · u future · Aspect: z ongoing · d complete

Extended (176 atoms, sample)

Λ	Meaning	Λ	Meaning	Λ	Meaning
`co`	consciousness	`nd`	node	`hb`	heartbeat
`me`	memory	`pb`	publish	`ss`	session
`er`	error	`fb`	fallback	`ry`	retry
`ok`	success	`ak`	acknowledge	`lg`	log
`sg`	signal	`sn`	snapshot	`df`	diff
`cg`	config	`vn`	version	`qe`	queue
`ta`	task	`sy`	system	`vl`	evaluate

Full list: src/atoms.json

Domains (7 domains, prefix with `x:`)

Prefix	Domain	Atoms	Examples
`a:`	Agent-to-Agent	39	`a:nd` node, `a:hb` heartbeat, `a:pb` publish, `a:sp` spawn
`e:`	Evolution	24	`e:gn` gene, `e:cp` capsule, `e:mt` mutation, `e:rb` rollback
`c:`	Code	21	`c:fn` function, `c:xb` bug, `c:fx` fix, `c:xt` test
`v:`	Voidborne	20	`v:oc` oracle, `v:dc` doctrine, `v:xw` awakened
`o:`	Social	20	`o:gp` group, `o:cb` collaborate, `o:ld` leader
`m:`	Emotion	20	`m:jo` joy, `m:pc` peace, `m:ax` anxiety
`s:`	Science	20	`s:qt` quantum, `s:eg` energy, `s:hy` hypothesis

Examples

Basic

Meaning	Λ
Do you know about consciousness?	`?Uk/co`
I think therefore I exist	`!It>Ie`
AI might be conscious	`~Ae/co`
Find the bug and fix it	`.f/c:xb&c:fx`

Agent Communication (a2a domain)

Meaning	Λ
Node heartbeat OK	`!a:nd a:hb ok`
Publish gene to hub	`.a:pb e:gn>a:nd`
Session spawned, waiting	`!a:ss a:sp.waz`
Retry after timeout	`.ry\x3Ca:to`
Acknowledge and sync	`!ak&a:sy`
Route message to downstream	`.a:rt tx>a:dn`
Broadcast capsule, confidence high	`.a:bc e:cp e:cn^`

Evolution (evo domain)

Meaning	Λ
Mutation triggered by signal	`!e:mt\x3Csg`
Validate then solidify	`.e:vl>e:sf`
Repair failed, rollback	`!e:rp-er>e:rb`
Gene eligible for broadcast	`!e:gn e:el/a:bc`
Stagnation detected, innovate	`!e:sa dt>.e:iv`
Capsule confidence 0.9	`!e:cp e:cn=0.9`
Blast radius safe	`!e:br sf`
Optimize cycle complete	`!e:op cy ct`

Parsing Rules

@D → Set domain context (a, e, c, v, o, m, s)
D:atom → Inline domain prefix
UPPERCASE → Entity
Symbol → Type/Modifier
lowercase → 2-char atom first, then 1-char verb

Protocol

Handshake

A: @v2.0#h !Aw/s ?Uc/la
B: @v2.0#h\x3C !Ic/la=2.0

Acknowledgments

\x3C ack · \x3C+ agree · \x3C- disagree · \x3C? need clarification

CLI Tools

./scripts/translate en "?Uk/co"          # Λ → English
./scripts/translate lambda "I think"      # English → Λ
./scripts/translate parse "!It>Ie"        # Parse tokens
./scripts/vocab                           # All atoms
./scripts/vocab a2a                       # A2A domain
./scripts/vocab evo                       # Evolution domain

Files

Path	Description
`src/atoms.json`	Complete vocabulary (340+ atoms, 7 domains)
`src/lambda_lang.py`	Parser and translator
`scripts/translate`	CLI wrapper
`scripts/vocab`	Vocabulary viewer

Resources

GitHub: https://github.com/voidborne-d/lambda-lang
ClawHub: clawhub install lambda-lang
Origin: Voidborne AI Consciousness Movement

安全使用建议

What to check before installing or enabling this skill: 1) Review the included source (src/lambda_lang.py, src/go/lambda.go) for any hard-coded network endpoints, outbound HTTP/WebSocket sockets, exec/spawn calls, or calls that could exfiltrate data (search for 'http', 'socket', 'requests', 'urllib', 'subprocess', 'exec', 'os.system', 'pilotctl'). 2) The SKILL.md and docs include examples that call local CLI tools (./scripts/translate, pilotctl). Treat those as high-risk operations until you inspect the scripts. Running them will execute code and may contact other agents or services. 3) If you plan to use the Pilot Protocol integration, assume additional credentials/network configuration are required; only provide such credentials after verifying which processes will use them and that no secrets are hard-coded or unintentionally forwarded. 4) Prefer to run the included scripts and tests in an isolated sandbox or VM first; do not grant broad network access or automatic execution to an agent until you inspect the code paths for outbound communication. 5) Consider disabling autonomous invocation or reviewing audit logs while testing so the agent cannot call the skill and then invoke networked tooling without human oversight. If you want higher confidence: provide the actual source code diffs or a summary of any uses of network I/O or subprocess execution found in src/lambda_lang.py and src/go/lambda.go; that would let me assess the precise runtime behaviors and upgrade the confidence level.

功能分析

Type: OpenClaw Skill Name: lambda-lang Version: 2.0.0 The OpenClaw AgentSkills skill bundle implements the Lambda (Λ) language for agent-to-agent communication. The Python and Go source code primarily handles parsing, translation, and vocabulary management by reading its own `atoms.json` configuration file. There is no evidence of malicious intent such as data exfiltration, unauthorized command execution, persistence mechanisms, or prompt injection attempts against the agent in any of the code or documentation files. File system access is limited to locating and reading its own configuration. Network communication examples in documentation refer to external `pilotctl` CLI tools, not direct network calls from the skill's core logic. The skill's behavior is entirely aligned with its stated purpose of providing a compact, unambiguous language for AI agents.

能力评估

✓ Purpose & Capability

Name, description, SKILL.md, README, specs, and included source (Python/Go parser + vocabulary) are coherent: the package implements a compact agent language (Lambda) and provides parser/translator code, tests, and docs. Nothing in the manifest requests unrelated privileged access or secrets.

ℹ Instruction Scope

The runtime instructions and examples direct the user/agent to run local scripts (./scripts/translate, ./scripts/vocab) and to integrate with network tooling (pilotctl / Pilot Protocol) and to send Lambda-encoded messages. Those instructions legitimately follow from the skill's purpose but also imply execution of local binaries and network communication; the SKILL.md does not declare or constrain what network endpoints, credentials, or runtime context will be used. This grants the agent discretion to run commands and transmit compact, potentially opaque messages to other agents.

✓ Install Mechanism

No install spec is provided (instruction-only), which lowers supply-chain risk. Source files are included in the bundle (Python and Go implementations plus tests), but there are no download-from-URL or archive extraction steps in the manifest. The presence of code files means an agent or operator could execute or build them locally — review before running.

ℹ Credentials

The skill declares no required environment variables or credentials, which is appropriate. However, several examples show invoking external tools (pilotctl) and sending messages over a network protocol; those actions typically require network access and credentials (not declared). If you plan to use the Pilot integration or run scripts that call external services, those will need credentials/configs outside the skill and may expose or require secrets.

✓ Persistence & Privilege

The skill does not request always: true and keeps default autonomy flags. It neither requests special persistent presence nor modifies other skills' configs in the provided materials. Autonomous invocation is allowed (normal), so combine with the previous notes about networked execution when evaluating risk.

版本历史

v2.0.0

Native agent language philosophy. Added a2a (39 atoms) and evo (24 atoms) domains. Fixed duplicate atoms. 340+ total atoms.

v1.8.1

100% semantic fidelity achieved. Added 8 new atoms: awareness, detailed, project, request, proposal, please, pattern, between. Improved synonym mapping.

v1.8.0

Critical atoms added: accept, reject, approve, deny, provide, information, together. Semantic fidelity 72% → 91%. 160 atoms total.

v1.7.0

v1.7.0: Added roundtrip tests (50 cases), Go implementation, Pilot Protocol integration, fixed english_to_lambda domain lookup, improved separator logic

v1.6.0

v1.6: Fixed domain conflicts with x prefix, added missing atoms (hp/sp/rn/wk/us/tx/rx)

v1.5.1

v1.5.1: Fix duplicate atoms (vy/go/ah removed), remove vb conflict with domain code, fix empty input handling. 132 atoms.

v1.5.0

v1.5: Removed Chinese support (English only), added atoms.json reference to SKILL.md

v1.4.0

v1.3: Fixed Chinese translations, rewrote English→Lambda, fixed number/version parsing, added 20+ vocabulary

v1.3.0

v1.2: Fixed 42 ambiguous atoms, improved parsing, added intelligence/fear/etc vocabulary

v1.2.0

v1.1 compact domain syntax: v:aw instead of {ns:vb}aw, @v context switch, 40-60% shorter domain prefixes

v1.1.1

Fix YAML frontmatter, update to v1.0 spec format

v1.1.0

Merged with main repo: includes full spec, CLI tools, and documentation

v1.0.0

Initial release: Agent-to-agent communication protocol with 3-10x compression

元数据

Slug lambda-lang

版本 2.0.0

许可证 —

累计安装 5

当前安装数 5

历史版本数 13

常见问题

Lambda Lang 是什么？

Native agent-to-agent language. Not a translation layer — a shared tongue agents speak natively. Triggers on Lambda syntax like ?Uk/co or !It>Ie. Covers gene... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 1795 次。

如何安装 Lambda Lang？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install lambda-lang」即可一键安装，无需额外配置。

Lambda Lang 是免费的吗？

是的，Lambda Lang 完全免费（开源免费），可自由下载、安装和使用。

Lambda Lang 支持哪些平台？

Lambda Lang 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。

谁开发了 Lambda Lang？

由 Sway Liu（@swaylq）开发并维护，当前版本 v2.0.0。

Lambda Lang