docs: add Atomic Agents, open-connector and DeepTutor integration guides #82
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,128 @@ | ||
| --- | ||
| title: "Atomic Agents" | ||
| icon: "/integration-logo/atomic-agents-icondoc.svg" | ||
| description: "Use Atomic Agents with Eden AI to build agents on 500+ AI models through an OpenAI-compatible, Instructor-powered interface." | ||
| --- | ||
|
|
||
| import { TechArticleSchema } from "/snippets/TechArticleSchema.mdx"; | ||
|
|
||
| <TechArticleSchema | ||
| title={"Atomic Agents"} | ||
| description={"Use Atomic Agents with Eden AI to build agents on 500+ AI models through an OpenAI-compatible, Instructor-powered interface."} | ||
| path="v3/integrations/atomic-agents" | ||
| articleSection="AI Frameworks" | ||
| about={"Agent Framework Integration"} | ||
| proficiencyLevel="Intermediate" | ||
| keywords={["Eden AI", "AI API", "Atomic Agents", "Python", "Instructor"]} | ||
| datePublished="2026-07-27T00:00:00Z" | ||
| dateModified="2026-07-27T00:00:00Z" | ||
| /> | ||
|
|
||
| Use Atomic Agents with Eden AI to build agents on 500+ AI models through an OpenAI-compatible, Instructor-powered interface. | ||
|
|
||
| ## Overview | ||
|
|
||
| [Atomic Agents](https://github.com/Eigenwise/atomic-agents) is a lightweight, modular Python framework for building AI agents on composable, schema-driven building blocks (via [Instructor](https://github.com/instructor-ai/instructor) and Pydantic). Atomic Agents works with Eden AI out of the box through its existing OpenAI-compatible client, so your agents reach models from OpenAI, Anthropic, Google, Mistral and more behind one key, with EU-based, GDPR-aligned inference. | ||
|
Check warning on line 25 in v3/integrations/atomic-agents.mdx
|
||
|
|
||
| ## Installation | ||
|
|
||
| Eden AI reuses the OpenAI client and Instructor that Atomic Agents already relies on: | ||
|
|
||
| <CodeGroup> | ||
| ```bash pip | ||
| pip install atomic-agents openai instructor | ||
| ``` | ||
|
|
||
| ```bash poetry | ||
| poetry add atomic-agents openai instructor | ||
| ``` | ||
| </CodeGroup> | ||
|
|
||
| ## Quick Start | ||
|
|
||
| Wrap an OpenAI client pointed at Eden AI with Instructor — the same pattern used for any other OpenAI-compatible provider — and pass it to your agent: | ||
|
|
||
| <CodeGroup> | ||
| ```bash .env | ||
| EDENAI_API_KEY=your_api_key_here | ||
| ``` | ||
|
|
||
| {/* skip-test */} | ||
| ```python Python | ||
| import os | ||
| import instructor | ||
| from openai import OpenAI | ||
| from atomic_agents.agents.base_agent import BaseAgent, BaseAgentConfig | ||
|
|
||
| client = instructor.from_openai( | ||
| OpenAI( | ||
| base_url="https://api.edenai.run/v3", | ||
| api_key=os.getenv("EDENAI_API_KEY"), | ||
| ) | ||
| ) | ||
|
|
||
| agent = BaseAgent( | ||
| config=BaseAgentConfig( | ||
| client=client, | ||
| model="anthropic/claude-sonnet-5", # <provider>/<model> | ||
| ) | ||
| ) | ||
|
|
||
| response = agent.run(agent.input_schema(chat_message="Hello! How are you?")) | ||
| print(response.chat_message) | ||
|
Comment on lines
+51
to
+72
There was a problem hiding this comment. 🎯 Functional Correctness | 🟠 Major | ⚡ Quick win 🧩 Analysis chain🏁 Script executed: #!/bin/bash
python -m pip install -U atomic-agents openai instructor
python - <<'PY'
from inspect import signature
from atomic_agents import AtomicAgent, AgentConfig, BasicChatInputSchema
assert "config" in signature(AtomicAgent).parameters
assert "chat_message" in signature(BasicChatInputSchema).parameters
PYRepository: edenai/docs Length of output: 312 🌐 Web query:
💡 Result: Atomic Agents is a lightweight, modular Python framework designed for building agentic AI pipelines and applications [1][2]. It leverages Pydantic for data validation and schema serialization, and is built on top of the Instructor library to support various LLM providers (e.g., OpenAI, Anthropic, Gemini, Ollama) [1][3][2]. Key Components and Definitions: AtomicAgent The AtomicAgent class is the core component for constructing AI agents [4][5]. It manages chat interactions, history, system prompts, and responses from language models [4][5]. It is instantiated with a configuration object and expects both an input schema and an output schema (typically via generic type parameters: AtomicAgent[InputSchema, OutputSchema]) [6][7]. AgentConfig AgentConfig is a Pydantic model that provides the configuration for an AtomicAgent [4][8]. It includes essential parameters such as the Instructor client, the LLM model identifier, optional history management (ChatHistory), and a system prompt generator [4][8]. As of recent updates (v2.0+), it no longer requires the explicit inclusion of input or output schemas, which are now handled at the agent class level [7]. BasicChatInputSchema BasicChatInputSchema is a standard input schema that inherits from BaseIOSchema [8][7]. It provides a default structure for user input in chat-based agents, typically containing a chat_message field [8]. BaseAgentConfig (Legacy) BaseAgentConfig is a deprecated class [7]. In earlier versions of the framework, it served the role now filled by AgentConfig [7]. Users upgrading from older versions must migrate to AgentConfig [7]. For reference, the framework also includes several other key classes such as ChatHistory (for memory management) and BaseIOSchema (the base class for all input/output schemas) [4][9][7]. The framework's architecture was significantly updated in version 2.0, with deprecated legacy names (like BaseAgent, BaseAgentConfig, etc.) being replaced by the current structures mentioned above [7]. [4][1][8][7] Citations:
🌐 Web query:
💡 Result: To get started with Atomic Agents, you can clone the main repository using the following command: git clone https://github.com/eigenwise/atomic-agents.git The Atomic Agents framework is designed to be a lightweight and modular system for building AI agents, built on top of Instructor and Pydantic [1][2]. Regarding the use of BaseAgent and InputSchema in your quickstart, Atomic Agents primarily utilizes the AtomicAgent class, which takes input and output schemas defined as Pydantic models [3][4]. Below is a standard pattern for creating an agent with custom schemas: from pydantic import Field from atomic_agents import AtomicAgent, AgentConfig, BaseIOSchema # 1. Define your schemas inheriting from BaseIOSchema class MyInputSchema(BaseIOSchema): query: str = Field(..., description="The user's query.") class MyOutputSchema(BaseIOSchema): response: str = Field(..., description="The agent's response.") # 2. Configure and initialize the agent agent = AtomicAgent[MyInputSchema, MyOutputSchema]( config=AgentConfig( client=your_instructor_client, model="gpt-4o", # or your preferred model #... other config options like system_prompt_generator or history)) # 3. Run the agent input_data = MyInputSchema(query="Hello, agent!") result = agent.run(input_data) print(result.response) Key resources for your quickstart include: - Official Documentation: https://eigenwise.github.io/atomic-agents/ [1] - GitHub Repository: https://github.com/eigenwise/atomic-agents [5] - Example directory: Check the atomic-examples folder in the repository for various implementations, including basic chatbots and custom schemas [5][2]. Note: Ensure you have installed the core library via pip install atomic-agents and any specific provider SDKs (e.g., openai, anthropic) required for your chosen model [1][6]. Citations:
🌐 Web query:
💡 Result: In the Atomic Agents framework, BasicChatInputSchema is the default input schema used for standard agent interactions [1][2]. It is defined as a Pydantic model that inherits from BaseIOSchema and contains a single required field, chat_message, which captures the user's input [1][2]. To use it in a quickstart context, you typically instantiate an AtomicAgent with BasicChatInputSchema (and a corresponding output schema, such as BasicChatOutputSchema) as type parameters, then pass an instance of the schema to the agent's run method [3][4]. Quickstart Example: import instructor import openai from atomic_agents import AtomicAgent, AgentConfig, BasicChatInputSchema, BasicChatOutputSchema from atomic_agents.context import ChatHistory # 1. Setup client = instructor.from_openai(openai.OpenAI) history = ChatHistory # 2. Initialize Agent agent = AtomicAgent[BasicChatInputSchema, BasicChatOutputSchema]( config=AgentConfig( client=client, model="gpt-5-mini", history=history)) # 3. Use BasicChatInputSchema to send a message user_message = "Tell me about quantum computing" response = agent.run(BasicChatInputSchema(chat_message=user_message)) print(response.chat_message) Key Implementation Details: - The schema is designed for simple, chat-based interactions [1][2]. - You can access the agent's expected input schema via agent.input_schema [5]. - For more complex interactions, you can define custom schemas inheriting from BaseIOSchema and replace BasicChatInputSchema in the agent definition [6]. Citations:
🌐 Web query:
💡 Result: In the atomic-agents library, the classes BaseAgent and BaseAgentConfig were deprecated and subsequently removed in version 2.0.0 as part of a major refactor [1][2]. The following replacements should be used: - BaseAgent has been renamed to AtomicAgent [3][1]. - BaseAgentConfig has been renamed to AgentConfig [3][1]. Additionally, the library underwent significant structural changes in version 2.0.0, including: - Simplification of import paths by eliminating the.lib directory [3][2]. - Renaming of other related classes, such as BaseAgentInputSchema to BasicChatInputSchema and AgentMemory to ChatHistory [1]. - Updates to how schemas are specified; for example, they are now typically handled via generic type parameters (e.g., AtomicAgent[InputSchema, OutputSchema]), and input_schema and output_schema fields have been removed from the configuration object [1]. Users migrating to v2.0.0 or later should consult the official UPGRADE_DOC.md in the repository for a complete list of breaking changes and migration steps [1]. Citations:
🏁 Script executed: #!/bin/bash
set -euo pipefail
# Locate and inspect the referenced atomic agents documentation snippet(s).
wd="$(pwd)"
if [ -z "$wd" ]; then
printf 'PWD unavailable\n'
exit 0
fi
printf 'Tracked candidates for atomic-agents.mdx:\n'
git ls-files | grep -E '(^|/)atomic-agents\.mdx$|atomic-agents' || true
printf '\nFile excerpt for v3/integrations/atomic-agents.mdx:\n'
if [ -f v3/integrations/atomic-agents.mdx ]; then
wc -l v3/integrations/atomic-agents.mdx
sed -n '1,130p' v3/integrations/atomic-agents.mdx | cat -n
else
echo 'v3/integrations/atomic-agents.mdx not found'
fi
printf '\nSearch for BaseAgent/BaseAgentConfig/input_schema usage:\n'
rg -n "BaseAgent|BaseAgentConfig|input_schema\\(|AtomicAgent|AgentConfig|BasicChatInputSchema|atomic-agents" v3/integrations/atomic-agents.mdx 2>/dev/null || trueRepository: edenai/docs Length of output: 6230 Update the Atomic Agents snippets to the current API. The installed 🤖 Prompt for AI Agents |
||
| ``` | ||
| </CodeGroup> | ||
|
|
||
| Both Instructor `TOOLS` and `JSON` modes work unchanged with Eden AI. | ||
|
|
||
| ## Switching models | ||
|
|
||
| Change the `model` string on `BaseAgentConfig` to reach any Eden AI model — no per-provider SDKs, no code changes: | ||
|
|
||
| <CodeGroup> | ||
| {/* skip-test */} | ||
| ```python Python | ||
| for model in [ | ||
| "openai/gpt-5.5", | ||
| "anthropic/claude-sonnet-5", | ||
| "mistral/mistral-large-2512", | ||
| ]: | ||
| agent = BaseAgent(config=BaseAgentConfig(client=client, model=model)) | ||
| response = agent.run(agent.input_schema(chat_message="Write a haiku about the sea.")) | ||
| print(model, "->", response.chat_message) | ||
| ``` | ||
| </CodeGroup> | ||
|
|
||
| ## Available Models | ||
|
|
||
| Use the `provider/model` format for any Eden AI model: | ||
|
|
||
| **OpenAI** | ||
| - `openai/gpt-5.5` | ||
| - `openai/gpt-5-mini` | ||
|
|
||
| **Anthropic** | ||
| - `anthropic/claude-sonnet-5` | ||
| - `anthropic/claude-haiku-4-5` | ||
|
|
||
| **Google** | ||
| - `google/gemini-2.5-pro` | ||
| - `google/gemini-3.5-flash` | ||
|
|
||
| **Mistral** | ||
| - `mistral/mistral-large-2512` | ||
| - `mistral/mistral-small-2603` | ||
|
|
||
| ## Environment Variables | ||
|
|
||
| <CodeGroup> | ||
| ```bash .env | ||
| EDENAI_API_KEY=your_api_key_here | ||
| ``` | ||
| </CodeGroup> | ||
|
|
||
| ## Next Steps | ||
|
|
||
| - [aisuite](/v3/integrations/aisuite) - Another OpenAI-style multi-provider interface | ||
| - [Chat Completions](/v3/llms/chat-completions) - Core LLM endpoint | ||
| - [List LLM Models](/v3/llms/listing-models) - Browse available providers and models | ||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,94 @@ | ||
| --- | ||
| title: "DeepTutor" | ||
| icon: "/integration-logo/deeptutor-icondoc.svg" | ||
| description: "Use DeepTutor with Eden AI to run your self-hosted study and research app on 500+ AI models through one API." | ||
| --- | ||
|
|
||
| import { TechArticleSchema } from "/snippets/TechArticleSchema.mdx"; | ||
|
|
||
| <TechArticleSchema | ||
| title={"DeepTutor"} | ||
| description={"Use DeepTutor with Eden AI to run your self-hosted study and research app on 500+ AI models through one API."} | ||
| path="v3/integrations/deeptutor" | ||
| articleSection="Chat UI" | ||
| about={"AI Study & Research App Integration"} | ||
| proficiencyLevel="Beginner" | ||
| keywords={["Eden AI", "AI API", "DeepTutor", "RAG", "Study"]} | ||
| datePublished="2026-07-27T00:00:00Z" | ||
| dateModified="2026-07-27T00:00:00Z" | ||
| /> | ||
|
|
||
| Use DeepTutor with Eden AI to run your self-hosted study and research app on 500+ AI models through one API. | ||
|
|
||
| ## Overview | ||
|
|
||
| [DeepTutor](https://github.com/HKUDS/DeepTutor) is a self-hosted AI study and research app — chat, quizzes, and RAG knowledge bases. Eden AI is registered as a **built-in gateway provider** in DeepTutor's model registry, so your whole instance can run on models from OpenAI, Anthropic, Google, Mistral and more behind one key, with EU-based, GDPR-aligned inference. | ||
|
|
||
| ## Installation | ||
|
|
||
| <CodeGroup> | ||
| ```bash pip | ||
| mkdir -p my-deeptutor && cd my-deeptutor | ||
| pip install -U deeptutor | ||
| ``` | ||
| </CodeGroup> | ||
|
|
||
| ## Quick Start | ||
|
|
||
| Select Eden AI as the LLM provider when the setup wizard prompts for one: | ||
|
|
||
| <CodeGroup> | ||
| ```bash Terminal | ||
| deeptutor init # prompts for LLM provider / base URL / API key / model | ||
| deeptutor start # starts backend + frontend | ||
| ``` | ||
| </CodeGroup> | ||
|
|
||
| Open the frontend URL printed in the terminal (default `http://127.0.0.1:3782`). Already running an instance? Add or change the provider later from **Settings → Models** in the web UI instead of re-running `init` — DeepTutor auto-detects Eden AI from the base URL. | ||
|
|
||
| ## Configuration | ||
|
|
||
| DeepTutor's provider registry stores the profile in `data/user/settings/model_catalog.json`: | ||
|
|
||
| | Field | Value | | ||
| |---|---| | ||
| | Provider | `edenai` | | ||
| | Base URL | `https://api.edenai.run/v3` | | ||
| | API key | `EDENAI_API_KEY` | | ||
| | Model | `provider/model`, e.g. `anthropic/claude-sonnet-5` | | ||
|
Comment on lines
+51
to
+58
There was a problem hiding this comment. 🎯 Functional Correctness | 🟠 Major | ⚡ Quick win 🧩 Analysis chain🏁 Script executed: #!/bin/bash
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/HKUDS/DeepTutor.git "$tmp/DeepTutor"
rg -n 'project-root.*\.env|model_catalog|Settings.*Models|dotenv' "$tmp/DeepTutor/README.md"Repository: edenai/docs Length of output: 2012 🏁 Script executed: #!/bin/bash
set -euo pipefail
# Inspect the relevant v3/deeptutor MDX sections and search the docs repo for DeepTutor EDENAI references.
if [ -f v3/integrations/deeptutor.mdx ]; then
echo "== v3/integrations/deeptutor.mdx relevant lines =="
sed -n '40,95p' v3/integrations/deeptutor.mdx | nl -ba -v40
else
echo "missing v3/integrations/deeptutor.mdx"
fi
echo "== .env/EDENAI/model_catalog mentions =="
rg -n 'EDENAI_API_KEY|Environment Variables|model_catalog\.json|deep.*tutor|deeptutor|DeepTutor|\.env' .Repository: edenai/docs Length of output: 240 🏁 Script executed: #!/bin/bash
set -euo pipefail
pwd
printf '\n== file exists? ==\n'
test -f v3/integrations/deeptutor.mdx && echo present || echo missing
printf '\n== v3/integrations/deeptutor.mdx relevant lines =="\n'
if test -f v3/integrations/deeptutor.mdx; then
sed -n '40,95p' v3/integrations/deeptutor.mdx | awk '{printf "%5d %s\n", NR+39, $0}'
fi
printf '\n== deep tutor/env/model catalog mentions in current repo =="\n'
rg -n 'EDENAI_API_KEY|Environment Variables|model_catalog\.json|deep.*tutor|deeptutor|DeepTutor|\.env' . || trueRepository: edenai/docs Length of output: 13674 Remove the unused DeepTutor configures the EDTN AI key through 🤖 Prompt for AI AgentsSource: Coding guidelines |
||
|
|
||
| ## Available Models | ||
|
|
||
| Use the `provider/model` format for any Eden AI model: | ||
|
|
||
| **OpenAI** | ||
| - `openai/gpt-5.5` | ||
| - `openai/gpt-5-mini` | ||
|
|
||
| **Anthropic** | ||
| - `anthropic/claude-sonnet-5` | ||
| - `anthropic/claude-haiku-4-5` | ||
|
|
||
| **Google** | ||
| - `google/gemini-2.5-pro` | ||
| - `google/gemini-3.5-flash` | ||
|
|
||
| **Mistral** | ||
| - `mistral/mistral-large-2512` | ||
| - `mistral/mistral-small-2603` | ||
|
|
||
| ## Environment Variables | ||
|
|
||
| <CodeGroup> | ||
| ```bash .env | ||
| EDENAI_API_KEY=your_api_key_here | ||
| ``` | ||
| </CodeGroup> | ||
|
|
||
| DeepTutor reads this key itself once the Eden AI provider is configured — it does not need to be exported manually in most setups (the Settings UI writes it into `model_catalog.json`). | ||
|
|
||
| ## Next Steps | ||
|
|
||
| - [Open WebUI](/v3/integrations/open-webui) - Another self-hosted chat interface | ||
| - [Chat Completions](/v3/llms/chat-completions) - Core LLM endpoint | ||
| - [List LLM Models](/v3/llms/listing-models) - Browse available providers and models | ||
| Original file line number | Diff line number | Diff line change | ||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| @@ -0,0 +1,100 @@ | ||||||||||||||||||||||||||
| --- | ||||||||||||||||||||||||||
| title: "open-connector" | ||||||||||||||||||||||||||
| icon: "/integration-logo/open-connector-icondoc.svg" | ||||||||||||||||||||||||||
| description: "Use open-connector with Eden AI to expose 500+ AI models through a self-hosted gateway with SDK, CLI, MCP and HTTP access." | ||||||||||||||||||||||||||
| --- | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| import { TechArticleSchema } from "/snippets/TechArticleSchema.mdx"; | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| <TechArticleSchema | ||||||||||||||||||||||||||
| title={"open-connector"} | ||||||||||||||||||||||||||
| description={"Use open-connector with Eden AI to expose 500+ AI models through a self-hosted gateway with SDK, CLI, MCP and HTTP access."} | ||||||||||||||||||||||||||
| path="v3/integrations/open-connector" | ||||||||||||||||||||||||||
| articleSection="AI Gateway" | ||||||||||||||||||||||||||
| about={"LLM Gateway Integration"} | ||||||||||||||||||||||||||
| proficiencyLevel="Intermediate" | ||||||||||||||||||||||||||
| keywords={["Eden AI", "AI API", "open-connector", "LLM Gateway", "MCP"]} | ||||||||||||||||||||||||||
| datePublished="2026-07-27T00:00:00Z" | ||||||||||||||||||||||||||
| dateModified="2026-07-27T00:00:00Z" | ||||||||||||||||||||||||||
| /> | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| Use open-connector with Eden AI to expose 500+ AI models through a self-hosted gateway with SDK, CLI, MCP and HTTP access. | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| ## Overview | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| [open-connector](https://github.com/oomol-lab/open-connector) is an open-source gateway connecting 1000+ SaaS providers to AI agents via SDK, CLI, MCP and HTTP. Eden AI ships as a **built-in connector**, so any agent wired through open-connector reaches models from OpenAI, Anthropic, Google, Mistral and more behind one key, with EU-based, GDPR-aligned inference and per-request cost visibility. | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| ## Installation | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| open-connector is a self-hosted gateway, distributed as a Docker image: | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| <CodeGroup> | ||||||||||||||||||||||||||
| ```bash Docker | ||||||||||||||||||||||||||
| docker compose up | ||||||||||||||||||||||||||
| ``` | ||||||||||||||||||||||||||
| </CodeGroup> | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| This pulls `ghcr.io/oomol-lab/open-connector:latest` and serves the console and API at `http://localhost:3000`. | ||||||||||||||||||||||||||
|
Comment on lines
+29
to
+37
There was a problem hiding this comment. 🎯 Functional Correctness | 🟠 Major | ⚡ Quick win 🧩 Analysis chain🏁 Script executed: #!/bin/bash
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/oomol-lab/open-connector.git "$tmp/open-connector"
test -f "$tmp/open-connector/docker-compose.yml"Repository: edenai/docs Length of output: 202 🏁 Script executed: #!/bin/bash
set -e
echo "== target file section =="
sed -n '1,80p' v3/integrations/open-connector.mdx | cat -n
echo
echo "== repository files mentioning open-connector docker compose =="
rg -n "open-connector|docker-compose|compose" v3/integrations/open-connector.mdx . -g '!node_modules' -g '!dist' || trueRepository: edenai/docs Length of output: 10015 Make the Docker quick start runnable from a clean checkout.
Proposed fix+git clone --depth 1 https://github.com/oomol-lab/open-connector.git
+cd open-connector
docker compose up📝 Committable suggestion
Suggested change
🤖 Prompt for AI Agents |
||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| ## Quick Start | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| Store your Eden AI key as a connection credential, then call the `edenai` service: | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| <CodeGroup> | ||||||||||||||||||||||||||
| ```bash Add credential | ||||||||||||||||||||||||||
| curl -s -X PUT http://localhost:3000/api/connections/edenai \ | ||||||||||||||||||||||||||
| -H 'content-type: application/json' \ | ||||||||||||||||||||||||||
| -d '{"authType":"api_key","values":{"apiKey":"your_api_key_here"}}' | ||||||||||||||||||||||||||
| ``` | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| ```bash Call Eden AI | ||||||||||||||||||||||||||
| curl -s -X POST http://localhost:3000/v1/actions/edenai.create_chat_completion \ | ||||||||||||||||||||||||||
| -H 'content-type: application/json' \ | ||||||||||||||||||||||||||
| -d '{ | ||||||||||||||||||||||||||
| "input": { | ||||||||||||||||||||||||||
| "model": "openai/gpt-5.5", | ||||||||||||||||||||||||||
| "messages": [{"role": "user", "content": "Hello! How are you?"}] | ||||||||||||||||||||||||||
| } | ||||||||||||||||||||||||||
| }' | ||||||||||||||||||||||||||
|
Comment on lines
+41
to
+58
There was a problem hiding this comment. 🔒 Security & Privacy | 🟠 Major | ⚡ Quick win 🧩 Analysis chain🏁 Script executed: #!/bin/bash
curl -fsS http://localhost:3000/openapi.json | jq '.components.securitySchemes'
curl -fsS http://localhost:3000/api/providers/edenai | jq .Repository: edenai/docs Length of output: 247 🏁 Script executed: #!/bin/bash
set -euo pipefail
echo "== repo files =="
git ls-files | sed -n '1,120p'
echo "== target file excerpt =="
if [ -f v3/integrations/open-connector.mdx ]; then
nl -ba v3/integrations/open-connector.mdx | sed -n '1,140p'
else
fd -a 'open-connector\.mdx' . | sed -n '1,20p'
fi
echo "== docs/credentials.md excerpt if present =="
if [ -f docs/credentials.md ]; then
nl -ba docs/credentials.md | sed -n '1,240p'
else
fd -a 'credentials\.md' . | sed -n '1,20p'
fi
echo "== mentions of OOMOL_CONNECT_ENCRYPTION_KEY =="
rg -n "OOMOL_CONNECT_ENCRYPTION_KEY|encryption|credential|api_key|authType|Authorization:" . -g '!node_modules' -g '!dist' -g '!build' | sed -n '1,260p'Repository: edenai/docs Length of output: 4296 🏁 Script executed: #!/bin/bash
set -euo pipefail
echo "== target file excerpt =="
if [ -f v3/integrations/open-connector.mdx ]; then
cat -n v3/integrations/open-connector.mdx | sed -n '1,160p'
else
echo "v3/integrations/open-connector.mdx not found"
fd -a 'open-connector\.mdx' . 2>/dev/null || true
fi
echo "== docs/credentials.md excerpt if present =="
if [ -f docs/credentials.md ]; then
cat -n docs/credentials.md | sed -n '1,260p'
else
fd -a 'credentials\.md' . 2>/dev/null | while read -r f; do echo "--- $f"; cat -n "$f" | sed -n '1,120p'; done
fi
echo "== repository-wide mentions =="
rg -n "OOMOL_CONNECT_ENCRYPTION_KEY|encryption|credential|authType|Authorization: Bearer|Bearer <api_key>|Bearer"| || true
echo "== api connection/provider/action examples in v3 integrations =="
cat -n v3/integrations/open-connector.mdx | sed -n '1,160p' | rg -n "connections|edenai|api/|v1/actions|Authorization|bearer|secret|auth" -C 3 || trueRepository: edenai/docs Length of output: 4467 🏁 Script executed: #!/bin/bash
set -euo pipefail
echo "== repository-wide relevant mentions =="
rg -n "OOMOL_CONNECT_ENCRYPTION_KEY|encryption|credential|authType|Authorization: Bearer|Bearer <api_key>|Bearer" . || true
echo "== API docs endpoint examples =="
rg -n "Authorization: Bearer <api_key>|Bearer <api_key>|security|Bearer" api-reference v3 -g '*.mdx' -g '*.json' | sed -n '1,120p'Repository: edenai/docs Length of output: 29506 🌐 Web query:
💡 Result: In OOMOL OpenConnector, the OOMOL_CONNECT_ENCRYPTION_KEY environment variable is used to enable AES-256-GCM encryption for sensitive data stored in the runtime database, such as SQLite [1][2]. When this key is configured, the system encrypts provider credentials, OAuth client configuration, and completed idempotent Action response payloads at rest [1][2]. If OOMOL_CONNECT_ENCRYPTION_KEY is not set, these items are stored in plaintext within the SQLite database [1][3]. In such cases, the runtime will issue a startup warning, and you should treat the SQLite database file as a sensitive, unprotected store [1][4]. Key management and usage details include: - External Storage: The OpenConnector runtime does not store the encryption key itself [1][3]. You are responsible for managing it securely using a secret manager or vault [3][5]. - Irrecoverability: If the encryption key is lost, any records encrypted with that key cannot be recovered [1][5]. - Key Rotation: You can rotate the encryption key by using the runtime:data rotate-key command, providing both the current (OOMOL_CONNECT_ENCRYPTION_KEY) and the new (OOMOL_CONNECT_NEW_ENCRYPTION_KEY) keys [1][6]. - Metadata: Even with encryption enabled, certain data remains unencrypted as metadata, including idempotency key hashes, request fingerprints, claim states, and timestamps [1][2]. Citations:
Document OpenConnector auth and credential-at-rest safeguards. The example omits the required bearer auth for the 🤖 Prompt for AI Agents |
||||||||||||||||||||||||||
| ``` | ||||||||||||||||||||||||||
| </CodeGroup> | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| The base URL (`https://api.edenai.run/v3`) is built into the connector — no extra configuration needed. The response includes the standard OpenAI-compatible fields plus a top-level `cost` field with the Eden AI billed cost. | ||||||||||||||||||||||||||
|
There was a problem hiding this comment. 🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win 🧩 Analysis chain🏁 Script executed: #!/bin/bash
curl -fsS http://localhost:3000/v1/actions/edenai.create_chat_completion \
-H "Authorization: Bearer ${OOMOL_CONNECT_RUNTIME_TOKEN}" \
-H 'content-type: application/json' \
-d '{"input":{"model":"openai/gpt-5.5","messages":[{"role":"user","content":"Hello"}]}}' |
jq '{top_level_cost: .cost, nested_cost: .data.cost, nested_choices: .data.choices}'Repository: edenai/docs Length of output: 198 🏁 Script executed: #!/bin/bash
set -euo pipefail
echo "== file context =="
sed -n '1,140p' v3/integrations/open-connector.mdx 2>/dev/null || true
echo
echo "== search open-connector/docs runtime api references =="
rg -n "data.cost|choices|success|message|data|meta|open-connector|edenai.create_chat_completion|create_chat_completion|runtime-api" . --glob '!node_modules' --glob '!dist' --glob '!build' --glob '!coverage' | head -200Repository: edenai/docs Length of output: 24432 🌐 Web query:
💡 Result: In the oomol-lab/open-connector project, the HTTP Runtime API (endpoints under /v1) utilizes a uniform JSON response envelope for its communication [1][2]. When a request is successful, the server returns a JSON response containing specific fields, including data and meta [3][4]. Based on the SDK implementation and test cases, the standard structure of this successful response envelope typically includes: data: The actual payload or result of the action or API call [3][5][6][4]. message: A human-readable status message (often "OK" for successful operations) [5][6][4]. meta: An object containing metadata regarding the request execution, which commonly includes [5][6][4]: - executionId: A server-assigned identifier for the specific execution [5][6][4]. - actionId: The identifier of the action being performed [5][6][4]. The SDK's executeRaw method explicitly maps these envelope fields into a RawResult object [6][4]. In contrast, the standard execute method returns the data field directly to the caller [5][4]. If a request fails, the server emits a different structure, generally Citations:
🏁 Script executed: #!/bin/bash
set -euo pipefail
python3 - <<'PY'
import urllib.request
urls = [
"https://raw.githubusercontent.com/oomol-lab/open-connector/main/docs/runtime-api.md",
"https://raw.githubusercontent.com/oomol-lab/open-connector/main/docs/runtime-api.md",
]
for url in urls[:1]:
with urllib.request.urlopen(url, timeout=20) as r:
text = r.read().decode("utf-8")
start = text.find("## Runtime API")
end = text.find("## SDK", start)
section = text[start:end] if end != -1 else text[start:start+1600]
print(section)
PY
echo
echo "== fetch open-connector repo runtime-api markdown excerpt =="
curl -fsS https://raw.githubusercontent.com/oomol-lab/open-connector/main/docs/runtime-api.md | sed -n '1,260p'Repository: edenai/docs Length of output: 2435 🌐 Web query:
💡 Result: The path /v1/actions/edenai.create_chat_completion refers to a specific Action within the OpenConnector ecosystem, which is an open-source gateway for connecting AI agents to various SaaS providers [1][2]. Regarding "data choices" and "cost": 1. Data Choices: OpenConnector acts as a mediator for API calls [3]. When using an action like edenai.create_chat_completion, the runtime handles the routing of your request to the Eden AI API [4][5]. The specific "data choices"—such as the model selection, message history, and other parameters—are defined by the schema of the Eden AI Chat Completions API [6][7]. OpenConnector exposes these through its generated OpenAPI specification, allowing you to inspect the required and optional inputs (e.g., model, messages, temperature) directly from the runtime's generated /openapi.json or through the Web Console [1][4]. 2. Cost: OpenConnector itself is an open-source gateway software [1][2] and does not inherently charge for executing actions. However, the cost of calling edenai.create_chat_completion depends entirely on your subscription or usage plan with Eden AI [7][8]. Because OpenConnector simply proxies these requests to the provider [4][5], any financial costs, rate limits, or data usage fees are governed by your direct agreement with Eden AI [7]. For technical implementation, you can call this action using the Connector SDK, the oo CLI, or by making a direct HTTP POST request to the /v1/actions/edenai.create_chat_completion endpoint on your OpenConnector runtime [1][9][4]. Ensure your runtime is configured with the necessary Eden AI credentials (usually an API key) in the Web Console, as the runtime maintains credential isolation [4][2][10]. Citations:
Describe the OpenConnector response envelope correctly. OpenConnector 🤖 Prompt for AI Agents |
||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| ## Available Models | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| The `edenai` service exposes `edenai.list_models` and `edenai.create_chat_completion`. Use the `provider/model` format for any Eden AI model: | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| **OpenAI** | ||||||||||||||||||||||||||
| - `openai/gpt-5.5` | ||||||||||||||||||||||||||
| - `openai/gpt-5-mini` | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| **Anthropic** | ||||||||||||||||||||||||||
| - `anthropic/claude-sonnet-5` | ||||||||||||||||||||||||||
| - `anthropic/claude-haiku-4-5` | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| **Google** | ||||||||||||||||||||||||||
| - `google/gemini-2.5-pro` | ||||||||||||||||||||||||||
| - `google/gemini-3.5-flash` | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| **Mistral** | ||||||||||||||||||||||||||
| - `mistral/mistral-large-2512` | ||||||||||||||||||||||||||
| - `mistral/mistral-small-2603` | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| ## Credentials | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| Unlike most integrations, open-connector stores the Eden AI key as a **connection credential** through its API rather than an environment variable: | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| <CodeGroup> | ||||||||||||||||||||||||||
| ```bash Update credential | ||||||||||||||||||||||||||
| curl -s -X PUT http://localhost:3000/api/connections/edenai \ | ||||||||||||||||||||||||||
| -H 'content-type: application/json' \ | ||||||||||||||||||||||||||
| -d '{"authType":"api_key","values":{"apiKey":"your_api_key_here"}}' | ||||||||||||||||||||||||||
| ``` | ||||||||||||||||||||||||||
| </CodeGroup> | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| ## Next Steps | ||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||
| - [Lynkr](/v3/integrations/lynkr) - Another self-hosted LLM gateway | ||||||||||||||||||||||||||
| - [Bifrost](/v3/integrations/bifrost) - Custom-provider LLM gateway | ||||||||||||||||||||||||||
| - [Chat Completions](/v3/llms/chat-completions) - Core LLM endpoint | ||||||||||||||||||||||||||
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Use consistent token placeholders across the integration guides.
These examples use
your_api_key_herewithout distinguishing production credentials from sandbox credentials.v3/integrations/atomic-agents.mdx#L46-L48: useapi_tokenfor the production.envexample.v3/integrations/atomic-agents.mdx#L119-L121: apply the same production/sandbox token naming.v3/integrations/open-connector.mdx#L44-L47: replace the JSON placeholder with an explicitapi_tokenorsandbox_api_token.v3/integrations/open-connector.mdx#L89-L92: apply the same token naming to the update example.As per coding guidelines, V3 examples must distinguish production
api_tokenfrom testingsandbox_api_token.📍 Affects 2 files
v3/integrations/atomic-agents.mdx#L46-L48(this comment)v3/integrations/atomic-agents.mdx#L119-L121v3/integrations/open-connector.mdx#L44-L47v3/integrations/open-connector.mdx#L89-L92🤖 Prompt for AI Agents
Source: Coding guidelines