candidate-kb
Manage candidate knowledge base. Use when ingesting CVs, adding profile information, or querying candidate details for resume/cover letter generation.
Incident response for Speak API outages: triage, fallback to offline
---
name: speak-incident-runbook
description: 'Incident response for Speak API outages: triage, fallback to offline
mode, and recovery procedures.
Use when implementing incident runbook,
or managing Speak language learning platform operations.
Trigger with phrases like "speak incident runbook", "speak incident runbook".
'
allowed-tools: Read, Write, Edit, Bash(npm:*), Bash(curl:*), Grep
version: 1.14.0
license: MIT
author: Jeremy Longshore <jeremy@intentsolutions.io>
tags:
- saas
- speak
- api
compatibility: Designed for Claude Code
---
# Speak Incident Runbook
## Overview
Incident response for Speak API outages: triage, fallback to offline mode, and recovery procedures.
## Prerequisites
- Completed `speak-install-auth` setup
- Valid API credentials configured
- Understanding of Speak API patterns
## Instructions
### Step 1: Configuration
Configure incident runbook for your Speak integration. Speak uses OpenAI's GPT-4o for AI tutoring and Whisper for speech recognition.
```typescript
// speak_incident_runbook_config.ts
const config = {
apiKey: process.env.SPEAK_API_KEY!,
appId: process.env.SPEAK_APP_ID!,
environment: process.env.NODE_ENV || 'development',
};
```
### Step 2: Implementation
```typescript
// Core implementation for speak incident runbook
import { SpeakClient } from '@speak/language-sdk';
const client = new SpeakClient(config);
// CI test with mocked responses
async function runCITests() {
const mockClient = new MockSpeakClient();
await mockClient.assessPronunciation({ audioPath: "test.wav", targetText: "hello", language: "en" });
console.log("CI tests passed");
}
```
### Step 3: Verification
```bash
npm test
```
## Output
- Speak Incident Runbook configured and verified
- CI pipeline with mocked Speak API tests
- Error handling and monitoring in place
## Error Handling
| Error | Cause | Solution |
|-------|-------|----------|
| 401 Unauthorized | Invalid API key | Verify SPEAK_API_KEY |
| 429 Rate Limited | Too many requests | Implement backoff |
| Connection timeout | Network issue | Check connectivity to api.speak.com |
| Audio format error | Wrong codec | Convert to WAV 16kHz mono |
## Resources
- [Speak Website](https://speak.com)
- [OpenAI Realtime API](https://platform.openai.com/docs/guides/realtime)
- [Speak GPT-4 Blog](https://speak.com/blog/speak-gpt-4)
## Next Steps
For deployment, see `speak-deploy-integration`.
## Examples
**Basic**: Apply incident runbook with default settings for a standard Speak integration.
**Production**: Configure with monitoring, alerting, and team-specific language learning requirements.
---
name: speak-incident-runbook
description: 'Incident response for Speak API outages: triage, fallback to offline
mode, and recovery procedures.
Use when implementing incident runbook,
or managing Speak language learning platform operations.
Trigger with phrases like "speak incident runbook", "speak incident runbook".
'
allowed-tools: Read, Write, Edit, Bash(npm:*), Bash(curl:*), Grep
version: 1.14.0
license: MIT
author: Jeremy Longshore <jeremy@intentsolutions.io>
tags:
- saas
- speak
- api
compatibility: Designed for Claude Code
---
# Speak Incident Runbook
## Overview
Incident response for Speak API outages: triage, fallback to offline mode, and recovery procedures.
## Prerequisites
- Completed `speak-install-auth` setup
- Valid API credentials configured
- Understanding of Speak API patterns
## Instructions
### Step 1: Configuration
Configure incident runbook for your Speak integration. Speak uses OpenAI's GPT-4o for AI tutoring and Whisper for speech recognition.
```typescript
// speak_incident_runbook_config.ts
const config = {
apiKey: process.env.SPEAK_API_KEY!,
appId: process.env.SPEAK_APP_ID!,
environment: process.env.NODE_ENV || 'development',
};
```
### Step 2: Implementation
```typescript
// Core implementation for speak incident runbook
import { SpeakClient } from '@speak/language-sdk';
const client = new SpeakClient(config);
// CI test with mocked responses
async function runCITests() {
const mockClient = new MockSpeakClient();
await mockClient.assessPronunciation({ audioPath: "test.wav", targetText: "hello", language: "en" });
console.log("CI tests passed");
}
```
### Step 3: Verification
```bash
npm test
```
## Output
- Speak Incident Runbook configured and verified
- CI pipeline with mocked Speak API tests
- Error handling and monitoring in place
## Error Handling
| Error | Cause | Solution |
|-------|-------|----------|
| 401 Unauthorized | Invalid API key | Verify SPEAK_API_KEY |
| 429 Rate Limited | Too many requests | Implement backoff |
| Connection timeout | Network issue | Check connectivity to api.speak.com |
| Audio format error | Wrong codec | Convert to WAV 16kHz mono |
## Resources
- [Speak Website](https://speak.com)
- [OpenAI Realtime API](https://platform.openai.com/docs/guides/realtime)
- [Speak GPT-4 Blog](https://speak.com/blog/speak-gpt-4)
## Next Steps
For deployment, see `speak-deploy-integration`.
## Examples
**Basic**: Apply incident runbook with default settings for a standard Speak integration.
**Production**: Configure with monitoring, alerting, and team-specific language learning requirements.
Imported from the listed AgentSkill.sh category. Complete supplied package files and notices are retained. Required external tools and credentials are not installed. No bundled code was executed during import. Individual agent task behavior has not been tested.