EpidBot OpenClaw Skill

Enables AI agents to interact with EpidBot's REST API for analyzing Brazilian public health data. EpidBot uses natural language processing to help users query, download, and analyze health data from DATASUS.

Overview

EpidBot provides access to:

- Brazilian health data: SINAN disease notifications, SIM mortality, SIH hospital admissions
International data sources: WHO, PAHO, HealthData.gov, World Bank, ECDC
Data analysis: Temporal trends, spatial distribution, demographic breakdowns
Visualizations: Charts, maps, heatmaps, and reports
SQL queries: Execute DuckDB SQL on parquet files

Authentication

Option 1: API Key (Recommended for agents)

1. Login to EpidBot web interface at https://epidbot.kwar-ai.com.br
Go to Admin Panel -> API Keys -> Create new API key
Set the API key as an environment variable:

CODEBLOCK0

Option 2: Username/Password (Returns JWT tokens)

CODEBLOCK1

API Base URL

Default: INLINECODE0

Configure via EPIDBOT_BASE_URL environment variable.

Quick Examples

Check API Health

CODEBLOCK2

Send a Chat Message (Async Submit + Poll)

Chat messages are processed asynchronously. Submit a message to get a job_id, then poll for the result.

CODEBLOCK3

List Available Tools

CODEBLOCK4

Tools / Capabilities

chat (async submit)

Submit a chat message for async processing. Returns a job_id immediately. LLM responses typically take 5-120 seconds.

Request:
CODEBLOCK5

Submit Response (200):
CODEBLOCK6

chat_poll

Poll for the status and result of a chat job. Endpoint: INLINECODE2

Recommended polling interval: 2-5 seconds.

Poll Response -- still processing:
CODEBLOCK7

Poll Response -- completed:
CODEBLOCK8

Poll Response -- failed:
CODEBLOCK9

chat_stream (WebSocket)

For streaming responses, connect via WebSocket at:
CODEBLOCK10

Client -> Server Messages:

CODEBLOCK11

Server -> Client Messages:

CODEBLOCK12

list_sessions

List all chat sessions for the authenticated user.

Endpoint: INLINECODE3

Output:
CODEBLOCK13

getsessionmessages

Get message history for a specific session.

Endpoint: INLINECODE4

Output:
CODEBLOCK14

list_reports

List all generated reports.

Endpoint: INLINECODE5

Output:
CODEBLOCK15

get_report

Get full details of a specific report.

Endpoint: INLINECODE6

Output:
CODEBLOCK16

download_report

Download report as a markdown file.

Endpoint: INLINECODE7

Output: File download with Content-Disposition header.

list_tools

List all available agent tools with descriptions.

Endpoint: INLINECODE9

Output:
CODEBLOCK17

Error Handling

All endpoints may return errors:

CODEBLOCK18

Common Status Codes:

- 400 - Bad Request (invalid input)
INLINECODE11 - Unauthorized (missing/invalid authentication)
INLINECODE12 - Forbidden (valid auth but insufficient permissions)
INLINECODE13 - Not Found
INLINECODE14 - Too Many Requests
INLINECODE15 - Internal Server Error

Agent Usage Patterns

Pattern 1: Async Chat (Submit + Poll) -- Recommended

CODEBLOCK19

Pattern 2: Streaming Chat (WebSocket)

CODEBLOCK20

Pattern 3: Session-Aware Chat

CODEBLOCK21

Pattern 4: Tool-Based Workflow

CODEBLOCK22

Available Data Sources

Brazilian Data (PySUS)

- SINAN: Disease notifications (dengue, Zika, chikungunya, measles, etc.)
SIM: Mortality data
SIH: Hospital admissions
CNES: Health facilities
IBGE: Census and demographic data

International Data

- WHO/GHO: Global health indicators
PAHO: Americas health data
HealthData.gov: US hospital capacity, COVID metrics
World Bank: Development indicators
ECDC: European disease surveillance

Specialized

- Mosqlimate/Infodengue: Brazilian epidemiological parameters

Configuration

Environment Variable	Default	Description
INLINECODE16	-	API key for authentication
INLINECODE17

https://api.epidbot.kwar-ai.com.br | Base URL of EpidBot API |

Rate Limits

Endpoint	Limit
INLINECODE19	10 requests/minute
INLINECODE20

5 requests/minute | | Other endpoints | 60 requests/minute |

Deployment

EpidBot is typically deployed via Docker Compose:

CODEBLOCK23

Health check: INLINECODE21

Limitations

- Job expiry: Chat jobs expire after 1 hour
File sizes: Large data exports may be limited by memory constraints
Sandbox execution: Python/SQL code execution happens in an isolated sandbox with resource limits

Support

- GitHub: https://github.com/fccoelho/EpiDBot
Documentation: https://github.com/fccoelho/EpiDBot/tree/main/docs

EpidBot OpenClaw 技能

使AI代理能够与EpidBot的REST API交互，用于分析巴西公共卫生数据。EpidBot使用自然语言处理帮助用户查询、下载和分析来自DATASUS的健康数据。

概述

EpidBot提供以下访问权限：

- 巴西健康数据：SINAN疾病通报、SIM死亡率、SIH住院记录
国际数据源：WHO、PAHO、HealthData.gov、世界银行、ECDC
数据分析：时间趋势、空间分布、人口统计细分
可视化：图表、地图、热力图和报告
SQL查询：对parquet文件执行DuckDB SQL

身份认证

选项1：API密钥（推荐用于代理）

1. 登录EpidBot网页界面 https://epidbot.kwar-ai.com.br
进入管理面板 -> API密钥 -> 创建新API密钥
将API密钥设置为环境变量：

bash
export EPIDBOTAPIKEY=your-api-key-here
export EPIDBOTBASEURL=https://api.epidbot.kwar-ai.com.br

选项2：用户名/密码（返回JWT令牌）

bash
curl -X POST $EPIDBOTBASEURL/api/v1/auth/login \
-H Content-Type: application/json \
-d {username: user, password: pass}

响应：

{

access_token: eyJ...,

refresh_token: eyJ...,

token_type: bearer,

expires_in: 900

}

API基础URL

默认值：https://api.epidbot.kwar-ai.com.br/api/v1

通过EPIDBOTBASEURL环境变量配置。

快速示例

检查API健康状态

bash
curl -H X-API-Key: $EPIDBOTAPIKEY \
$EPIDBOTBASEURL/api/v1/health

发送聊天消息（异步提交 + 轮询）

聊天消息异步处理。提交消息获取job_id，然后轮询获取结果。

bash

步骤1：提交消息（立即返回job_id）

JOB=$(curl -s -X POST $EPIDBOTBASEURL/api/v1/chat \
-H X-API-Key: $EPIDBOTAPIKEY \
-H Content-Type: application/json \
-d {message: 显示2023年圣保罗的登革热病例, locale: zh})

JOBID=$(echo $JOB | jq -r .jobid)

步骤2：轮询结果（每3秒）

while true; do RESULT=$(curl -s $EPIDBOTBASEURL/api/v1/chat/$JOB_ID \ -H X-API-Key: $EPIDBOTAPIKEY) STATUS=$(echo $RESULT | jq -r .status) if [ $STATUS = completed ] || [ $STATUS = failed ]; then echo $RESULT | jq . break fi sleep 3 done

列出可用工具

bash
curl -H X-API-Key: $EPIDBOTAPIKEY \
$EPIDBOTBASEURL/api/v1/tools

工具/能力

chat（异步提交）

提交聊天消息进行异步处理。立即返回job_id。LLM响应通常需要5-120秒。

请求：
json
{
message: 2023年登革热有哪些可用数据？,
session_id: null,
locale: zh
}

提交响应（200）：
json
{
jobid: joba1b2c3d4...,
session_id: 1,
status: processing
}

chat_poll

轮询聊天任务的状态和结果。端点：GET /api/v1/chat/{job_id}

推荐轮询间隔：2-5秒。

轮询响应 -- 仍在处理：
json
{
jobid: joba1b2c3d4...,
status: processing,
created_at: 2024-01-01T00:00:00Z,
updated_at: 2024-01-01T00:00:05Z
}

轮询响应 -- 已完成：
json
{
jobid: joba1b2c3d4...,
status: completed,
session_id: 1,
content: EpidBot可以访问2023年的SINAN登革热数据...,
images: [],
thinking: 用户正在询问可用数据...,
created_at: 2024-01-01T00:00:00Z,
updated_at: 2024-01-01T00:00:45Z
}

轮询响应 -- 失败：
json
{
jobid: joba1b2c3d4...,
status: failed,
error: 错误消息描述,
created_at: 2024-01-01T00:00:00Z,
updated_at: 2024-01-01T00:00:10Z
}

chat_stream（WebSocket）

用于流式响应，通过WebSocket连接：

wss://api.epidbot.kwar-ai.com.br/api/v1/chat/stream?api_key=

客户端 -> 服务器消息：

json
{type: start, payload: {message: ..., session_id: null, locale: zh}}
{type: cancel}
{type: ping}

服务器 -> 客户端消息：

json
{type: thinking, data: {content: ...}}
{type: chunk, data: {content: ...}}
{type: complete, data: {content: ..., images: [], usage: {...}}}
{type: error, data: {error: ...}}
{type: cancelled}
{type: pong}

list_sessions

列出已认证用户的所有聊天会话。

端点： GET /api/v1/sessions

输出：
json
[
{
id: 1,
name: 登革热分析,
message_count: 12,
created_at: 2024-01-01T00:00:00Z,
updated_at: 2024-01-02T00:00:00Z
}
]

getsessionmessages

获取特定会话的消息历史。

端点： GET /api/v1/sessions/{session_id}/messages

输出：
json
{
session_id: 1,
messages: [
{
id: 1,
role: user,
text_content: 显示登革热数据,
thinking: null,
created_at: 2024-01-01T00:00:00Z
},
{
id: 2,
role: assistant,
text_content: 以下是登革热数据...,
thinking: 用户正在询问...,
created_at: 2024-01-01T00:00:01Z
}
]
}

list_reports

列出所有生成的报告。

端点： GET /api/v1/reports

输出：
json
[
{
id: 1,
title: 2023年登革热分析,
report_type: analysis,
image_count: 3,
contentsizebytes: 15234,
has_pdf: true,
created_at: 2024-01-01T00:00:00Z
}
]

get_report

获取特定报告的完整详情。

端点： GET /api/v1/reports/{report_id}

输出：
json
{
id: 1,
title: 2023年登革热分析,
report_type: analysis,
prompt: 显示登革热病例...,
content: # 登革热分析\n\n...,
image_count: 3,
contentsizebytes: 15234,
has_pdf: true,
created_at: 2024-01-01T00:00:00Z
}

download_report

下载报告为markdown文件。

端点： GET /api/v1/reports/{report_id}/download

输出： 文件下载，包含Content-Disposition头。

epidbotEpidBot交互

epidbot

EpidBot OpenClaw Skill

Overview

Authentication

Option 1: API Key (Recommended for agents)

Option 2: Username/Password (Returns JWT tokens)

API Base URL

Quick Examples

Check API Health

Send a Chat Message (Async Submit + Poll)

List Available Tools

Tools / Capabilities

chat (async submit)

chat_poll

chat_stream (WebSocket)

list_sessions

getsessionmessages

list_reports

get_report

download_report

list_tools

Error Handling

Agent Usage Patterns

Pattern 1: Async Chat (Submit + Poll) -- Recommended

Pattern 2: Streaming Chat (WebSocket)

Pattern 3: Session-Aware Chat

Pattern 4: Tool-Based Workflow

Available Data Sources

Brazilian Data (PySUS)

International Data

Specialized

Configuration

Rate Limits

Deployment

Limitations

Support

EpidBot OpenClaw 技能

概述

身份认证

选项1：API密钥（推荐用于代理）

选项2：用户名/密码（返回JWT令牌）

响应：

{

access_token: eyJ...,

refresh_token: eyJ...,

token_type: bearer,

expires_in: 900

}

API基础URL

快速示例

检查API健康状态

发送聊天消息（异步提交 + 轮询）

步骤1：提交消息（立即返回job_id）

步骤2：轮询结果（每3秒）

列出可用工具

工具/能力

chat（异步提交）

chat_poll

chat_stream（WebSocket）

list_sessions

getsessionmessages

list_reports

get_report

download_report

list_t

标签

通过对话安装

方式一：安装 SkillHub 和技能

方式二：设置 SkillHub 为优先技能安装源

通过命令行安装

下载

相关推荐

self-improvement

self-improvement

self-improvement

self-improvement