[draft] Code interpreter

[Weves]

Description

[Provide a brief description of the changes in this PR]

How Has This Been Tested?

[Describe the tests you ran to verify your changes]

Backporting (check the box to trigger backport action)

Note: You have to check that the action passes, otherwise resolve the conflicts manually and tag the patches.

• This PR should be backported (make sure to check that the backport attempt succeeds)
• [Optional] Override Linear Check

---

Summary by cubic

Adds a Code Interpreter tool that runs Python in an isolated service, stages input files, and returns stdout/stderr plus generated artifacts back into the chat. Includes server config, DB seeding, UI toggles, and tests.

• New Features

  • New AnalysisTool (run_code_interpreter) calling POST /execute with code, stdin, timeout, and staged files.
  • Persists generated files to the file store, infers mime types, and surfaces them in chat; text files are added to the next prompt.
  • Tool becomes a built-in; constructed only when CODE_INTERPRETER_BASE_URL is set.
  • Env config: CODE_INTERPRETER_BASE_URL, CODE_INTERPRETER_DEFAULT_TIMEOUT_MS (default 30000), CODE_INTERPRETER_REQUEST_TIMEOUT_SECONDS (default 30).
  • Admin UI: toggle “Code Interpreter”; tool shown with a terminal icon; setup hint when not configured.
  • Unit tests cover artifact persistence, lazy file loading, and service error handling.

• Migration

  • Run DB migrations to seed the tool entry.
  • Set CODE_INTERPRETER_BASE_URL (required), optionally adjust timeouts.
  • Restart the server, then enable “Code Interpreter” in the Assistant settings.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Preview	Comments	Updated (UTC)
internal-search	Ready	Preview	Comment	Sep 22, 2025 4:07am

[Weves]

6 Jobs Failed:

Run Integration Tests v2 / integration-tests (tests/migrations, tests-migrations) failed on "Run Integration Tests for tests-migrations"

[...]
            )
            tools = result.fetchall()
    
            # Should have all 5 builtin tools
>           assert (
                len(tools) == 5
            ), f"Should have created exactly 5 builtin tools, got {len(tools)}"
E           AssertionError: Should have created exactly 5 builtin tools, got 6
E           assert 6 == 5
E            +  where 6 = len([(1, 'SearchTool', 'Internal Search', 'The Search Action allows the Assistant to search through connected knowledge to... write artifacts, stream stdout and stderr, and return generated outputs for the chat session.', 'AnalysisTool', None)])

tests/integration/tests/migrations/test_tool_seeding.py:48: AssertionError
=========================== short test summary info ============================
FAILED tests/integration/tests/migrations/test_tool_seeding.py::test_tool_seeding_migration
=================== 1 failed, 2 passed, 5 skipped in 12.10s ====================
CRITICAL: Exiting due to uncaught exception AssertionError: Should have created exactly 5 builtin tools, got 6
assert 6 == 5
 +  where 6 = len([(1, 'SearchTool', 'Internal Search', 'The Search Action allows the Assistant to search through connected knowledge to... write artifacts, stream stdout and stderr, and return generated outputs for the chat session.', 'AnalysisTool', None)])
Error: Final attempt failed. Child_process exited with error code 1

Run Integration Tests v2 / required failed on "Run actions/github-script@v7"

[...]
  retry-exempt-status-codes: 400,401,403,404,422
env:
  PRIVATE_REGISTRY: experimental-registry.blacksmith.sh:5000
  PRIVATE_REGISTRY_USERNAME: ***
  PRIVATE_REGISTRY_PASSWORD: ***
  OPENAI_API_KEY: ***
  SLACK_BOT_TOKEN: ***
  CONFLUENCE_TEST_SPACE_URL: ***
  CONFLUENCE_USER_NAME: ***
  CONFLUENCE_ACCESS_TOKEN: ***
  JIRA_BASE_URL: ***
  JIRA_USER_EMAIL: ***
  JIRA_API_TOKEN: ***
  PERM_SYNC_SHAREPOINT_CLIENT_ID: ***
  PERM_SYNC_SHAREPOINT_PRIVATE_KEY: ***
  PERM_SYNC_SHAREPOINT_CERTIFICATE_PASSWORD: ***
  PERM_SYNC_SHAREPOINT_DIRECTORY_ID: ***
  GITHUB_REPO_NAME: ***/onyx
Error: One or more upstream jobs failed or were cancelled.

4 jobs failed running on non-Blacksmith runners.

---

Summary: 4 successful workflows, 5 failed workflows

• ✅ Quality Checks PR (1 job succeeded)
• ❌ Python Unit Tests (1 job failed)
• ❌ Run Integration Tests v2 (34 jobs succeeded, 2 jobs failed)
• ✅ Helm - Lint and Test Charts (1 job succeeded)
• ❌ Python Checks (1 job failed)
• ✅ External Dependency Unit Tests (5 jobs succeeded)
• ✅ Check Lazy Imports (1 job succeeded)
• ❌ Ensure PR references Linear (1 job failed)
• ❌ Connector Tests (1 job failed)

Last updated: 2025-09-22 04:25:48 UTC