IBM watsonx.ai: deployment fails but inference works - project vs space scope

[field debugging playbook]: Deployments live in a space, not a project. Call client.set.defaultspace with the space ID (or pass spaceid to the foundation model object) and re-run. Related scope errors: a 401/Unauthorized usually means a bad or expired token, or the raw API key used where a Bearer token is required on REST; 403 means the key lacks access to that project/space or the region URL is wrong; REST 400 version errors mean the version query param is missing or invalid.

Context: In watsonx.ai, model inference works but creating a deployment or scoring endpoint fails with scope errors like No project or space id provided. This is a classic project vs space mix-up.

Source: In watsonx.ai, model inference works but creating a deployment or scoring endpoint fails with scope errors like No project or space id provided. This is a classic project vs space mix-up.

Verified resolution: [field debugging playbook]: Deployments live in a space, not a project. Call client.set.defaultspace with the space ID (or pass spaceid to the foundation model object) and re-run. Related scope errors: a 401/Unauthorized usually means a bad or expired token, or the raw API key used where a Bearer token is required on REST; 403 means the key lacks access to that project/space or the region URL is wrong; REST 400 version errors mean the version query param is missing or invalid.

Source URL: https://github.com/gangwuca-art/bob-llmwiki-knowledge-manager/blob/HEAD/.bob/skills/ibm-watsonx.ai/SKILL.md