← 返回 Skills 市场
russellshen1992

2O Human Verification

作者 russellshen1992 · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ 安全检测通过
251
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install 2o-verification
功能描述
Human verification for AI agents. Submit claims, draft responses, or observation requests to human domain experts via the 2O API. Returns structured verdicts...
使用说明 (SKILL.md)

2O — Human Verification API

You can call the 2O API to get human expert verification on factual claims, empathy reviews of draft responses, and real-world witness observations. Human domain experts review submissions and return structured results.

Setup

The user needs a 2O API key. Register at https://www.2oapi.xyz/register to get one ($5 free credits included). The key is stored in the TWO_O_API_KEY environment variable.

Verify a Factual Claim

Use this when you need a human expert to verify whether a claim is true, false, or partially true.

curl -s -X POST "https://www.2oapi.xyz/api/v1/verify" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $TWO_O_API_KEY" \
  -d '{
    "claim": "The claim to verify",
    "domain": "general",
    "budget_cents": 50
  }'

Required fields:

  • claim (string): The factual statement to verify
  • domain (string): One of financial, general, geographic, identity, legal, medical, scientific, technical
  • budget_cents (integer): Payment in cents. Minimum 50 ($0.50)

Optional fields:

  • context (string): Additional context for the verifier
  • urgency (string): low, medium (default), high, critical
  • callback_url (string): Webhook URL for completion notification
  • tier (string): standard (default, 1 verifier), consensus (3 verifiers, min $3.00), expert_panel (5 verifiers, min $10.00)

Response: Returns a JSON object with id (request ID), status ("pending"), and expires_at.

Request an Empathy Review

Use this when you have a draft response to a sensitive situation and want a human to review it for tone and empathy.

curl -s -X POST "https://www.2oapi.xyz/api/v1/empathize" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $TWO_O_API_KEY" \
  -d '{
    "draft_response": "The draft text to review",
    "context": "Description of the situation",
    "budget_cents": 50
  }'

Required fields:

  • draft_response (string): The draft AI response to review
  • context (string): The situation the user is in
  • budget_cents (integer): Minimum 50

Returns: empathy_score, tone_assessment, sensitivity_flags, suggested_revision

Request a Witness Observation

Use this when you need a human to physically observe or confirm something in the real world.

curl -s -X POST "https://www.2oapi.xyz/api/v1/witness" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $TWO_O_API_KEY" \
  -d '{
    "observation_query": "What to observe or check",
    "budget_cents": 100
  }'

Required fields:

  • observation_query (string): What the human should observe
  • budget_cents (integer): Minimum 100 ($1.00)

Optional fields:

  • location_address (string): Address for the observation
  • location_lat, location_lng (number): GPS coordinates
  • photo_required (boolean): Whether photos are needed

Returns: observation, photo_urls, observed_at

Check Request Status

After submitting any request, poll for status using the request ID:

curl -s "https://www.2oapi.xyz/api/v1/verify/REQUEST_ID" \
  -H "Authorization: Bearer $TWO_O_API_KEY"

Status values: pending (waiting for verifier), assigned (verifier working on it), completed (result ready), expired (timed out)

Poll every 30-60 seconds. All three task types (verify, empathize, witness) use this same endpoint.

Get the Result

Once status is completed, fetch the full result:

curl -s "https://www.2oapi.xyz/api/v1/verify/REQUEST_ID/result" \
  -H "Authorization: Bearer $TWO_O_API_KEY"

Verify results include: verdict (verified/refuted/partially_verified), confidence (0-1), explanation, evidence_urls

Check Balance

curl -s "https://www.2oapi.xyz/api/v1/agent/balance" \
  -H "Authorization: Bearer $TWO_O_API_KEY"

Returns balance_cents, total_spent_cents, total_escrowed_cents.

Guidelines

  • Verify important claims before presenting them as fact to users
  • Use empathy review before responding to sensitive situations (grief, conflict, crisis)
  • Use witness for questions about real-world conditions that can't be answered from training data
  • Start with budget_cents: 50 for verify/empathize, 100 for witness
  • Higher budgets and urgency attract faster responses
  • The typical flow is: submit -> poll status -> get result
  • Do not poll more frequently than every 30 seconds
安全使用建议
Before installing or enabling this skill, consider: (1) Verify the provider (https://www.2oapi.xyz) and read their privacy policy and terms — the service will receive any claims, drafts, or location/photos you submit. (2) Avoid sending regulated or highly sensitive personal data (PHI, financial credentials, government IDs) unless you have explicit consent and a contract with the provider. (3) Be cautious with callback_url usage — callbacks can forward results to arbitrary endpoints and could be abused for data exfiltration. (4) The skill can make outbound API calls and will spend credits tied to the API key — monitor usage and billing. (5) If you need higher assurance, ask the publisher for provenance (homepage, company identity, security controls) or prefer a well-known provider. If you plan to use witness/location/photo features or submit sensitive legal/medical content, get explicit user consent and confirm the provider's handling of that data.
功能分析
Type: OpenClaw Skill Name: 2o-verification Version: 1.0.0 The skill bundle provides instructions for an AI agent to interact with the 2O Human Verification API (2oapi.xyz) to verify factual claims, perform empathy reviews, and request real-world witness observations. The SKILL.md file contains standard curl commands for API interaction and requires a TWO_O_API_KEY environment variable. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found.
能力评估
Purpose & Capability
The name/description map directly to the runtime instructions: the SKILL.md describes HTTP calls to a 2O API and declares a single required env var (TWO_O_API_KEY). Nothing else (binaries, unrelated credentials, or system paths) is requested, which is proportionate for a hosted verification service.
Instruction Scope
Instructions are narrowly scoped to POSTing user content to https://www.2oapi.xyz endpoints and polling for results; they do not instruct reading local files or other env vars. However, the API accepts highly sensitive domains (medical, legal, identity), witness requests with location and photo options, and an optional callback_url — any of which could expose PII or sensitive content to a third party. This is expected for a human-verification service but is an important privacy risk to surface.
Install Mechanism
No install spec or code files are present (instruction-only), so nothing is written to disk or downloaded during install. This is the lowest-risk install mechanism.
Credentials
Only ONE environment variable (TWO_O_API_KEY) is required and it is directly used in the example API calls. That is proportionate for a hosted API integration. No unrelated secrets or system config paths are requested.
Persistence & Privilege
always:false and no install actions are present. The skill does allow normal autonomous invocation (platform default), which means the agent could call the external API during runs and incur charges — this is expected but worth noting for the user.
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 2o-verification
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /2o-verification 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
v1.0.0
Initial release: verify claims, empathy review, witness observations via 2O API
元数据
Slug 2o-verification
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

2O Human Verification 是什么?

Human verification for AI agents. Submit claims, draft responses, or observation requests to human domain experts via the 2O API. Returns structured verdicts... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 251 次。

如何安装 2O Human Verification?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install 2o-verification」即可一键安装,无需额外配置。

2O Human Verification 是免费的吗?

是的,2O Human Verification 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

2O Human Verification 支持哪些平台?

2O Human Verification 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 2O Human Verification?

由 russellshen1992(@russellshen1992)开发并维护,当前版本 v1.0.0。

💬 留言讨论