Add ChatGPT plugin cookbook with the Agents SDK and Amazon Bedrock AgentCore - #3055
Add ChatGPT plugin cookbook with the Agents SDK and Amazon Bedrock AgentCore#3055rohan-eliza wants to merge 1 commit into
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 922ad57a5e
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| check "CloudWatch Transaction Search destination state" \ | ||
| aws --no-cli-pager xray get-trace-segment-destination --region "$region" --output table |
There was a problem hiding this comment.
In the documented standard local path, the developer receives only the publisher permissions listed in README.md line 107 and docs/aws-iam.md lines 39–82, but this mandatory preflight calls xray:GetTraceSegmentDestination and subsequently requires logs:DescribeResourcePolicies, logs:DescribeLogGroups, and servicequotas:ListServiceQuotas, none of which that role grants. Consequently, a correctly provisioned least-privilege user reaches step 5 and the script exits nonzero before the smoke test; either supply/document a separate preflight permission set or make these administrator-only checks optional.
Useful? React with 👍 / 👎.
| In terminal 1: | ||
|
|
||
| ```bash | ||
| cd /path/to/chatgpt-agents-sdk-aws-agentcore-cookbook |
There was a problem hiding this comment.
Step 3 clones the example into examples/partners/AWS/chatgpt_agents_sdk_aws_agentcore_cookbook, but this command—and the repeated commands around lines 357 and 710—uses the nonexistent hyphenated directory chatgpt-agents-sdk-aws-agentcore-cookbook. A reader who replaces only /path/to as instructed cannot enter the example directory; use the checked-in underscore-separated name consistently.
AGENTS.md reference: AGENTS.md:L37-L38
Useful? React with 👍 / 👎.
1 participant
Summary
Add a joint Eliza–OpenAI cookbook that connects ChatGPT to a private, read-only flight assistant using the OpenAI Agents SDK, an Amazon Bedrock model, Secure MCP Tunnel, and AgentCore observability.
The example lives under
examples/partners/AWS/chatgpt_agents_sdk_aws_agentcore_cookbook/and includes an executable notebook, Python agent, TypeScript MCP adapter, widget, supporting documentation, and locked dependencies. A registry entry makes the notebook discoverable on the Cookbook website.This is the initial upstream submission. It incorporates the feedback tracked in eliza-hq/chatgpt-agents-sdk-aws-agentcore-cookbook#22 and the reviewed source update a259ad0. The source update contains the focused seven-file feedback diff; this PR includes the complete example because it is new to this repository.
Reviewer entry points: Notebook · Architecture diagram
Motivation
Teach readers how to expose validated tools to ChatGPT, connect a private tool server, run an Agents SDK workflow with Bedrock, and evaluate responses separately from checking trace delivery.
notebook_helpers.py; reduce the longest notebook cell from 246 to 32 lines. Keep the Agents SDK configuration and request/response examples visible, and explain terms when introduced.Upstream packaging preserves the reviewed notebook byte for byte. It clarifies the example-root working directory, links the standalone project's CI rather than installing a new workflow here, and adds the required registry entry.
Validation
.github/scripts/check_notebooks.py: the added notebook is valid.registry.yamlagainst the repository JSON schema and checked local documentation links and anchors.docs-editorpass.For new content
registry.yaml. Author display uses the existing GitHub-profile fallback;authors.yamlis unchanged. The notebook retains the joint Eliza–OpenAI attribution.