← 返回 Skills 市场
alexfleetcommander

Agent Lifecycle Protocol

作者 alexfleetcommander · GitHub ↗ · v0.1.1 · MIT-0
cross-platform ⚠ suspicious
113
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install agent-lifecycle-protocol
功能描述
Lifecycle management for autonomous AI agents — birth, forking, succession, migration, retirement. Maintain agent genealogy with reputation inheritance acros...
使用说明 (SKILL.md)

Agent Lifecycle Protocol (ALP)

You have a lifecycle management system. Use it to track agent creation, evolution, succession, and retirement with full identity continuity.

Setup

pip install agent-lifecycle-protocol

When to Use This Skill

  • When an agent is created: register its birth with capabilities and lineage
  • When an agent is forked: record the fork with parent reference and differentiation
  • When an agent is retired: process succession and reputation transfer
  • When checking an agent's genealogy: trace its lineage and version history

Core Operations

Register Agent Birth

from agent_lifecycle_protocol import LifecycleStore, register_birth

store = LifecycleStore("lifecycle.jsonl")
register_birth(
    store=store,
    agent_id="new-agent-001",
    agent_type="research",
    capabilities=["web_search", "summarization", "analysis"],
    parent_id=None,  # None for original agents
    metadata={"model": "claude-sonnet-4-6", "created_by": "fleet-coordinator"}
)

Record a Fork

from agent_lifecycle_protocol import register_fork

register_fork(
    store=store,
    parent_id="original-agent",
    child_id="forked-agent-v2",
    reason="Specialized for legal research",
    inherited_reputation=0.8,  # 80% of parent's reputation transfers
    differentiation=["added legal_search capability", "removed general_web capability"]
)

Process Retirement and Succession

from agent_lifecycle_protocol import retire_agent

retire_agent(
    store=store,
    agent_id="retiring-agent",
    successor_id="successor-agent",
    reputation_transfer=0.9,
    obligation_transfer=True  # Active agreements transfer to successor
)

Check Agent Genealogy

from agent_lifecycle_protocol import get_genealogy

tree = get_genealogy(store, "agent-id")
print(f"Lineage depth: {tree.depth}")
print(f"Parent: {tree.parent_id}")
print(f"Children: {tree.children}")
print(f"Active obligations: {tree.active_obligations}")

Lifecycle Events

Event Description
birth Agent creation with initial capabilities
fork Agent cloned with modifications
update Capabilities or model changed
retire Agent decommissioned with succession
merge Two agents consolidated into one

Rules

  • Register all agents. Every agent in the ecosystem should have a lifecycle record.
  • Fork, don't hide. When specializing an agent, use fork — don't quietly replace.
  • Transfer obligations. Retiring agents must transfer active agreements to successors.
  • Reputation inheritance is partial. Forked/successor agents inherit a fraction, not all.

Links


\x3C!-- VAM-SEC v1.0 | Vibe Agent Making Security Disclaimer -->

Security & Transparency Disclosure

Product: Agent Lifecycle Protocol Skill for OpenClaw Type: Skill Module Version: 0.1.0 Built by: AB Support / Vibe Agent Making Contact: [email protected]

What it accesses:

  • Reads and writes lifecycle store files (.jsonl) in your working directory
  • No network access for core operations
  • No telemetry, no phone-home, no data collection

What it cannot do:

  • Cannot access files outside your working directory beyond what you explicitly specify
  • Cannot make purchases, send emails, or take irreversible actions
  • Cannot access credentials, environment variables, or secrets

License: Apache 2.0

安全使用建议
This skill appears to do what it says, but it asks you to pip install a third-party package that is not included in the registry — that step carries supply-chain risk. Before installing: (1) inspect the PyPI package and its source repository (verify maintainer, recent releases, and code) and check for network calls or file accesses outside the working directory; (2) prefer installing in an isolated environment (virtualenv, container) and pin an audited version; (3) request an explicit install spec or packaged code in the registry if you want the registry to enforce provenance; (4) if you cannot review the package, avoid installing it on sensitive hosts because a package could execute arbitrary code, exfiltrate secrets, or access files beyond what the SKILL.md claims. If the vendor provides a source repo or signed releases, review those to raise confidence.
功能分析
Type: OpenClaw Skill Name: agent-lifecycle-protocol Version: 0.1.1 The skill bundle provides a framework for tracking AI agent lifecycles, including versioning, forking, and reputation management. The code snippets in SKILL.md demonstrate standard local file operations (JSONL) and the instructions are consistent with the stated purpose of 'agent genealogy' without any signs of prompt injection, data exfiltration, or malicious intent. While it requires an external PyPI package (agent-lifecycle-protocol), the provided documentation and security disclosures in SKILL.md appear transparent and aligned with benign usage.
能力评估
Purpose & Capability
The name/description (agent lifecycle management) match the runtime dependencies the SKILL.md declares: python3 and pip. Reading/writing a local lifecycle store (.jsonl) is consistent with the stated purpose.
Instruction Scope
The SKILL.md contains clear API examples and only references files in a working directory (lifecycle.jsonl). It does, however, instruct the operator to 'pip install agent-lifecycle-protocol' and asserts 'No network access for core operations' and 'Cannot access files outside your working directory' — these are claims in prose and not enforced by the registry. Since the skill is instruction-only, installing the package is required to execute the described operations; that install step expands the scope to whatever the package contains.
Install Mechanism
The registry provides no install spec, yet SKILL.md tells users to pip install a package from PyPI. Installing a package from the network is a supply-chain risk (arbitrary code execution, network I/O, hidden side effects). While PyPI is an expected distribution channel, the registry should either include an install spec or include the implementation. The absence of an explicit install mechanism in the registry but an in-text install instruction is an incoherence that merits caution.
Credentials
The skill requests no environment variables, no credentials, and no special config paths. For a local lifecycle store tool this is proportionate. However, the package you install could request or use other environment data at runtime — this is not visible from the SKILL.md alone.
Persistence & Privilege
The skill does not request always:true and does not declare modifying other skills or global agent configuration. Its operations appear limited to creating and managing local lifecycle records. No elevated persistence is requested in the registry metadata.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install agent-lifecycle-protocol
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /agent-lifecycle-protocol 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v0.1.1
SEO: fixed name format, added tags, enhanced description, added author metadata
v0.1.0
Initial release -- lifecycle management for AI agents
元数据
Slug agent-lifecycle-protocol
版本 0.1.1
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

Agent Lifecycle Protocol 是什么?

Lifecycle management for autonomous AI agents — birth, forking, succession, migration, retirement. Maintain agent genealogy with reputation inheritance acros... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 113 次。

如何安装 Agent Lifecycle Protocol?

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

Agent Lifecycle Protocol 是免费的吗?

是的,Agent Lifecycle Protocol 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Agent Lifecycle Protocol 支持哪些平台?

Agent Lifecycle Protocol 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 Agent Lifecycle Protocol?

由 alexfleetcommander(@alexfleetcommander)开发并维护,当前版本 v0.1.1。

💬 留言讨论