ai-services

Name: ai-services
Author: digitalocean-labs

15installs4forks

How to install ai-services?

digitalocean-labs/do-app-platform-skills/ai-services

$npx -y skills add digitalocean-labs/do-app-platform-skills --skill ai-services

Installs into the current project.

Summary

Configure DigitalOcean Gradient AI serverless inference and Agent Development Kit. Use when adding LLM inference, model access keys, serverless AI endpoints, or building AI agents with ADK on App Platform.

Files · 1

View on GitHub

SKILL.md

1	# AI Services Skill
2
3	Configure DigitalOcean Gradient AI Platform for App Platform applications.
4
5	> Tip: This is one specialized skill in the App Platform library. For complex multi-step projects, consider using the planner skill to generate a staged approach. For an overview of all available skills, see the [root SKILL.md](../../SKILL.md).
6
7	---
8
9	## Quick Decision
10
11	```
12	What do you need?
13	├── Simple LLM API calls → Serverless Inference
14	│ OpenAI-compatible API, no agent management
15	│
16	└── Full AI agents → Agent Development Kit (ADK)
17	Knowledge bases, RAG, guardrails, multi-agent routing
18	```
19
20	\| Need \| Solution \| Reference \|
21	\|------\|----------\|-----------\|
22	\| Call LLM models directly \| Serverless Inference \| [serverless-inference.md](reference/serverless-inference.md) \|
23	\| Build agents with knowledge bases \| ADK \| [agent-development-kit.md](reference/agent-development-kit.md) \|
24	\| Content filtering / guardrails \| ADK \| [agent-development-kit.md](reference/agent-development-kit.md) \|
25	\| Multi-agent workflows \| ADK \| [agent-development-kit.md](reference/agent-development-kit.md) \|
26
27	---
28
29	## Credential Handling
30
31	Model access keys follow the standard credential hierarchy:
32
33	1. GitHub Secrets (recommended): User creates key → adds to GitHub Secrets → app spec references
34	2. App Platform Secrets: Set via `doctl apps update` with `type: SECRET`
35
36	```yaml
37	# App Spec pattern
38	envs:
39	- key: MODEL_ACCESS_KEY
40	scope: RUN_TIME
41	type: SECRET
42	value: ${MODEL_ACCESS_KEY} # From GitHub Secrets
43	```
44
45	Key creation: Control Panel → Serverless Inference → Model Access Keys
46
47	> Keys shown only once after creation—store securely.
48
49	---
50
51	## Quick Start: Serverless Inference
52
53	```yaml
54	# .do/app.yaml
55	services:
56	- name: api
57	envs:
58	- key: MODEL_ACCESS_KEY
59	scope: RUN_TIME
60	type: SECRET
61	value: ${MODEL_ACCESS_KEY}
62	- key: INFERENCE_ENDPOINT
63	value: https://inference.do-ai.run
64	```
65
66	```python
67	# Python SDK (OpenAI-compatible)
68	from openai import OpenAI
69	import os
70
71	client = OpenAI(
72	base_url=os.environ["INFERENCE_ENDPOINT"] + "/v1",
73	api_key=os.environ["MODEL_ACCESS_KEY"],
74	)
75
76	response = client.chat.completions.create(
77	model="llama3.3-70b-instruct",
78	messages=[{"role": "user", "content": "Hello!"}],
79	)
80	```
81
82	Full guide: See [serverless-inference.md](reference/serverless-inference.md)
83
84	---
85
86	## Quick Start: Agent Development Kit
87
88	```bash
89	# Install and configure
90	pip install gradient-adk
91	gradient agent configure
92
93	# Run locally
94	gradient agent run
95	# → http://localhost:8080/run
96
97	# Deploy to DigitalOcean
98	gradient agent deploy
99	```
100
101	```python
102	# Agent entrypoint
103	from gradient_adk import entrypoint
104
105	@entrypoint
106	def entry(payload, context):
107	query = payload["prompt"]
108	return {"response": "Hello from agent!"}
109	```
110
111	Full guide: See [agent-development-kit.md](reference/agent-development-kit.md)
112
113	---
114
115	## Available Models
116
117	\| Model \| Use Case \|
118	\|-------\|----------\|
119	\| `llama3.3-70b-instruct` \| General purpose, high quality \|
120	\| `llama3-8b` \| Faster, lower cost \|
121	\| `mistral-7b` \| Efficient, multilingual \|
122
123	```bash
124	# List all available models
125	doctl genai list-models
126	```
127
128	Check [Gradient AI Models](https://docs.digitalocean.com/products/gradient-ai-platform/details/models/) for current availability.
129
130	---
131
132	## Reference Files
133
134	- [serverless-inference.md](reference/serverless-inference.md) — SDK setup, API parameters, examples
135	- [agent-development-kit.md](reference/agent-development-kit.md) — ADK workflow, knowledge bases, guardrails
136
137	---
138
139	## Quick Troubleshooting
140
141	\| Error \| Cause \| Fix \|
142	\|-------\|-------\|-----\|
143	\| `401 Unauthorized` \| Invalid model access key \| Verify key in GitHub Secrets \|
144	\| `Model not found` \| Invalid model ID \| Run `doctl genai list-models` \|
145	\| `Rate limit exceeded` \| Too many requests \| Implement exponential backoff \|
146	\| ADK deploy fails \| Missing token scopes \| Ensure `genai` CRUD + `project` read scopes \|
147
148	---
149
150	## Integration with Other Skills
151
152	- → designer: Add AI service environment variables to app spec
153	- → deployment: Model access key stored in GitHub Secrets
154	- → devcontainers: Test AI integrations locally before deployment
155	- → planner: Plan AI-enabled app deployments
156
157	---
158
159	## Documentation Links
160
161	- [Gradient AI Platform](https://docs.digitalocean.com/products/gradient-ai-platform/)
162	- [Available Models](https://docs.digitalocean.com/products/gradient-ai-platform/details/models/)
163	- [Serverless Inference](https://docs.digitalocean.com/products/gradient-ai-platform/how-to/serverless-inference/)
164	- [Agent Development Kit](https://docs.digitalocean.com/products/gradient-ai-platform/how-to/adk/)

People who liked this also viewed

6 picks

Type

ai-services

15installs4forks

View on GitHub

How to install ai-services?

digitalocean-labs/do-app-platform-skills/ai-services

$npx -y skills add digitalocean-labs/do-app-platform-skills --skill ai-services

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# AI Services Skill
2
3	Configure DigitalOcean Gradient AI Platform for App Platform applications.
4
5	> Tip: This is one specialized skill in the App Platform library. For complex multi-step projects, consider using the planner skill to generate a staged approach. For an overview of all available skills, see the [root SKILL.md](../../SKILL.md).
6
7	---
8
9	## Quick Decision
10
11	```
12	What do you need?
13	├── Simple LLM API calls → Serverless Inference
14	│ OpenAI-compatible API, no agent management
15	│
16	└── Full AI agents → Agent Development Kit (ADK)
17	Knowledge bases, RAG, guardrails, multi-agent routing
18	```
19
20	\| Need \| Solution \| Reference \|
21	\|------\|----------\|-----------\|
22	\| Call LLM models directly \| Serverless Inference \| [serverless-inference.md](reference/serverless-inference.md) \|
23	\| Build agents with knowledge bases \| ADK \| [agent-development-kit.md](reference/agent-development-kit.md) \|
24	\| Content filtering / guardrails \| ADK \| [agent-development-kit.md](reference/agent-development-kit.md) \|
25	\| Multi-agent workflows \| ADK \| [agent-development-kit.md](reference/agent-development-kit.md) \|
26
27	---
28
29	## Credential Handling
30
31	Model access keys follow the standard credential hierarchy:
32
33	1. GitHub Secrets (recommended): User creates key → adds to GitHub Secrets → app spec references
34	2. App Platform Secrets: Set via `doctl apps update` with `type: SECRET`
35
36	```yaml
37	# App Spec pattern
38	envs:
39	- key: MODEL_ACCESS_KEY
40	scope: RUN_TIME
41	type: SECRET
42	value: ${MODEL_ACCESS_KEY} # From GitHub Secrets
43	```
44
45	Key creation: Control Panel → Serverless Inference → Model Access Keys
46
47	> Keys shown only once after creation—store securely.
48
49	---
50
51	## Quick Start: Serverless Inference
52
53	```yaml
54	# .do/app.yaml
55	services:
56	- name: api
57	envs:
58	- key: MODEL_ACCESS_KEY
59	scope: RUN_TIME
60	type: SECRET
61	value: ${MODEL_ACCESS_KEY}
62	- key: INFERENCE_ENDPOINT
63	value: https://inference.do-ai.run
64	```
65
66	```python
67	# Python SDK (OpenAI-compatible)
68	from openai import OpenAI
69	import os
70
71	client = OpenAI(
72	base_url=os.environ["INFERENCE_ENDPOINT"] + "/v1",
73	api_key=os.environ["MODEL_ACCESS_KEY"],
74	)
75
76	response = client.chat.completions.create(
77	model="llama3.3-70b-instruct",
78	messages=[{"role": "user", "content": "Hello!"}],
79	)
80	```
81
82	Full guide: See [serverless-inference.md](reference/serverless-inference.md)
83
84	---
85
86	## Quick Start: Agent Development Kit
87
88	```bash
89	# Install and configure
90	pip install gradient-adk
91	gradient agent configure
92
93	# Run locally
94	gradient agent run
95	# → http://localhost:8080/run
96
97	# Deploy to DigitalOcean
98	gradient agent deploy
99	```
100
101	```python
102	# Agent entrypoint
103	from gradient_adk import entrypoint
104
105	@entrypoint
106	def entry(payload, context):
107	query = payload["prompt"]
108	return {"response": "Hello from agent!"}
109	```
110
111	Full guide: See [agent-development-kit.md](reference/agent-development-kit.md)
112
113	---
114
115	## Available Models
116
117	\| Model \| Use Case \|
118	\|-------\|----------\|
119	\| `llama3.3-70b-instruct` \| General purpose, high quality \|
120	\| `llama3-8b` \| Faster, lower cost \|
121	\| `mistral-7b` \| Efficient, multilingual \|
122
123	```bash
124	# List all available models
125	doctl genai list-models
126	```
127
128	Check [Gradient AI Models](https://docs.digitalocean.com/products/gradient-ai-platform/details/models/) for current availability.
129
130	---
131
132	## Reference Files
133
134	- [serverless-inference.md](reference/serverless-inference.md) — SDK setup, API parameters, examples
135	- [agent-development-kit.md](reference/agent-development-kit.md) — ADK workflow, knowledge bases, guardrails
136
137	---
138
139	## Quick Troubleshooting
140
141	\| Error \| Cause \| Fix \|
142	\|-------\|-------\|-----\|
143	\| `401 Unauthorized` \| Invalid model access key \| Verify key in GitHub Secrets \|
144	\| `Model not found` \| Invalid model ID \| Run `doctl genai list-models` \|
145	\| `Rate limit exceeded` \| Too many requests \| Implement exponential backoff \|
146	\| ADK deploy fails \| Missing token scopes \| Ensure `genai` CRUD + `project` read scopes \|
147
148	---
149
150	## Integration with Other Skills
151
152	- → designer: Add AI service environment variables to app spec
153	- → deployment: Model access key stored in GitHub Secrets
154	- → devcontainers: Test AI integrations locally before deployment
155	- → planner: Plan AI-enabled app deployments
156
157	---
158
159	## Documentation Links
160
161	- [Gradient AI Platform](https://docs.digitalocean.com/products/gradient-ai-platform/)
162	- [Available Models](https://docs.digitalocean.com/products/gradient-ai-platform/details/models/)
163	- [Serverless Inference](https://docs.digitalocean.com/products/gradient-ai-platform/how-to/serverless-inference/)
164	- [Agent Development Kit](https://docs.digitalocean.com/products/gradient-ai-platform/how-to/adk/)

People who liked this also viewed

6 picks

Type