dlthub-platform-profiles

name	dlthub-platform-profiles
description	ALWAYS read and follow this skill before acting. Profiles

Profiles

1. Locally pipelines and datasets run on dev profile.
2. On runtime pipelines (batch jobs) run on prod profile
3. On runtime notebooks (interactive jobs) run on access profile. if not defined they run on prod profile.
4. If user pins a different profile - it is used to run pipelines and datasets locally.
5. Get workspace info (via cli or mcp tool) to see a list of profiles. configured profiles are one

Secrets and configs

1. Profiles have corresponding secrets and configs.
2. config.toml and secrets.toml apply to all profiles. Keep only common settings in them
3. dev.config.toml and dev.secrets.toml define settings specific to dev profile
4. Settings in profile-scoped toml files overwrite workspace-scoped toml files.
5. Use dlthub ai secrets as usual — update-fragment requires --path, so always pass the right file! dlthub ai secrets list when in doubt.

Switch profile

• set WORKSPACE__PROFILE env variable when running Python commands to temporarily switch profile
• pin profile with dlthub local profile use <name> to change profile in this session. require to switch mcp server

Reference https://dlthub.com/docs/hub/core-concepts/profiles-dlthub.md