Skip to content

Latest commit

 

History

History

hash-ai-worker-ts

README.md

HASH AI Worker (TypeScript)

A Temporal worker for running AI-inference workflows.

Configuration

The service uses the following environment variables:

  • HASH_TEMPORAL_SERVER_HOST: The hostname that the Temporal server is running on (defaults to localhost).
  • HASH_TEMPORAL_SERVER_PORT: The port that the Temporal server is running on (defaults to 7233).
  • OPENAI_API_KEY: The OpenAI API key that is made available to workflows and activities.
  • ANTHROPIC_API_KEY: The Anthropic API key that is made available to workflows and activities.
  • HASH_GRAPH_API_HOST: The host address that the HASH Graph API is running on, e.g. graph, 127.0.0.1
  • HASH_GRAPH_API_PORT: The port that the HASH Graph API is running on, e.g. 4000
  • INTERNAL_API_HOST: The host for the internal API, required if the internal API is not running locally for workflows making use of the getWebSearchResultsActivity activity
  • INTERNAL_API_KEY: The API key used to authenticate with the internal API, required for workflows making use of the getWebSearchResultsActivity activity
  • HASH_VAULT_HOST: The host address (including protocol) that the Vault server is running on, e.g. http://127.0.0.1
  • HASH_VAULT_PORT: The port that the Vault server is running on, e.g. 8200
  • HASH_VAULT_ROOT_TOKEN: The token to authenticate with the Vault server.

Run the worker

  • Ensure the environment variables above are set, either in .env.local or in your shell.
  • Install dependencies:
    • yarn
  • Run the worker:
    • yarn dev

Logging

To help inspect the workings of the flow, logs of different levels of detail are written to different locations.

In development, LOG_LEVEL=debug is advised (and hardcoded into the yarn dev command for this app) for the most detailed logs.

Important to know:

  • If LOG_LEVEL=DEBUG, every LLM request will be logged.
  • In development and test only:
    • The console output for each Flow run is written to the file system in the activities/shared/flow-run-logs directory.
    • Each LLM request and response pair is written to the file system in activities/shared/get-llm-response/logs directory (assuming LOG_LEVEL=debug).
  • For LLM requests, detailed fields request and response are omitted from the console, but are available in the individual request files.
  • If DATADOG_API_KEY is set in the environment, logs will be sent to Datadog.

Running AI-dependent tests / optimization

To enable loading environment variables into tests, TEST_AI=true must be set in the environment.

e.g. to run a specific test, from this folder (hash-ai-worker-ts):

TEST_AI=true npx vitest get-entity-summaries-from-text.ai.test.ts

or

TEST_AI=true LOG_LEVEL=debug npx vitest get-entity-summaries-from-text.ai.test.ts