| name | deploy |
| description | Deploy the portfolio chat agent Lambda function to AWS. Use when the user asks to deploy, redeploy, push to Lambda, update the chat agent, or after updating context files or lambda_function.py. Rebuilds the knowledge base, packages dependencies, and deploys to AWS Lambda.
|
Deploy Skill
Rebuild and deploy the portfolio chat agent Lambda function to AWS.
When to Use
- After updating any file in
WebContent/context/
- After modifying
lambda/lambda_function.py
- When the user says "deploy", "redeploy", "push to Lambda", or "update the chat agent"
Workflow
- Rebuild knowledge base — Regenerate
lambda/knowledge_base.json from context files
- Clean previous build — Remove old
package/ directory and deployment.zip
- Install dependencies — Install Python packages for Linux/x86_64 target
- Package — Copy handler and knowledge base into package, create zip
- Deploy — Update Lambda function code via AWS CLI
- Verify — Confirm deployment succeeded
Commands
Step 1: Rebuild Knowledge Base
python scripts/build_knowledge_base.py
Verify output shows expected project count and no errors.
Step 2: Clean and Install Dependencies
cd lambda
rm -rf package/ deployment.zip
pip install -r requirements.txt -t package/ \
--platform manylinux2014_x86_64 \
--only-binary=:all: \
--implementation cp \
--python-version 3.12
The --platform and --only-binary flags ensure compiled extensions are built for Amazon Linux, not macOS.
Step 3: Package Deployment Zip
cp lambda_function.py knowledge_base.json package/
cd package && zip -r ../deployment.zip . && cd ..
Step 4: Deploy to Lambda
aws lambda update-function-code \
--function-name portfolio-chat-agent \
--zip-file fileb://lambda/deployment.zip
Run this from the repository root so the fileb:// path resolves correctly.
Step 5: Verify
Check that the response includes:
"LastUpdateStatus": "InProgress" or "Successful""FunctionName": "portfolio-chat-agent"- Updated
CodeSha256 value
Configuration
| Setting | Value |
|---|
| Function name | portfolio-chat-agent |
| Runtime | python3.12 |
| Region | us-west-1 |
| Memory | 256 MB |
| Timeout | 30 seconds |
| Architecture | x86_64 |
Rules
- Never print or log the Anthropic API key. If the AWS CLI response includes environment variables, do not echo them back to the user.
- Always rebuild the knowledge base first. Even if only
lambda_function.py changed, rebuilding ensures the knowledge base is current.
- Run from the repository root. The
fileb:// path in the deploy command is relative to the working directory.
- Do not modify Lambda environment variables unless the user explicitly asks to update the API key.
Troubleshooting
| Problem | Solution |
|---|
No module named 'pydantic_core._pydantic_core' | Missing --platform flag during pip install — dependencies were built for macOS instead of Linux |
Unable to load paramfile fileb:// | Wrong working directory — run the deploy command from the repository root |
ResourceNotFoundException | Lambda function doesn't exist yet — use aws lambda create-function (see docs/plans/chat-agent/aws-setup.md Step 7) |
CodeStorageExceededException | Deployment zip too large — check for unnecessary dependencies in requirements.txt |
Reference Documentation
- aws-setup.md — full AWS account setup, IAM roles, initial Lambda creation, and Function URL configuration
