Check a new internal Marin developer’s local environment and access to development services. Distinguish setup problems from missing permissions, repair authorized local issues, and provide a checklist of verified capabilities and remaining access requests.
Verify or complete a new internal Marin developer's local setup and access to GitHub, GCP, Iris, Weights & Biases, Hugging Face, and optional CoreWeave storage. Use when a team member asks to onboard, validate onboarding, or diagnose missing development access.
Run score Not tested
See how to installReview the instructions the agent will receive before you install.
Establish which parts of the developer environment are ready, fix local setup when safe, and identify each missing external grant. Keep secrets out of command output and the final report.
Use these as the current sources of truth:
docs/tutorials/installation.md.docs/dev-guide/contributing.md.lib/iris/OPS.md and the use-iris
skill.docs/tutorials/cloud-gpu.md.infra/pulumi/README.md, the
add-grant skill, and the review-grant skill.Do not copy procedures from infra/README.md; it is an infrastructure index.
Start with read-only checks. Report each area as ready, missing access, missing local setup, or not checked.
hai-gcp-models project, Application
Default Credentials, and read access to gs://marin-us-central2. Confirm the
current principal has projects/hai-gcp-models/roles/marindev with a filtered
IAM query that prints only the matching role name. Do not print credential
contents or the complete project policy.marin GCP Pulumi stack:
marindev check above, confirm the current principal
has projects/hai-gcp-models/roles/marinPulumiAdmin with the same filtered
IAM query.pulumi -C infra/pulumi stack export --stack marin >/dev/null to verify
state access without printing state contents.pulumi up as an onboarding check. Do not grant roles or mutate
live IAM unless the user asks for the grant workflow.iris login
only for an interactive human session; let the browser or headless login flow
request the human's input.WANDB_API_KEY and HF_TOKEN are present without printing
their values. When useful, perform a read-only identity check with the
service's CLI and confirm access to the marin-community W&B entity. Do not
persist a token outside the user's chosen credential store.Distinguish local configuration failures from permissions that a maintainer must grant. GCP project access, Iris IAP access, GitHub access, Weights & Biases, Hugging Face, Pulumi operator access, and CoreWeave storage are independent.
Run local import or CPU checks when they are cheap and do not download a large dataset. A remote Iris job changes shared state: submit one only when the user explicitly authorizes the smoke job. Keep it CPU-only, small, and bounded. Never request a GPU or TPU during onboarding.
Do not start, stop, restart, deploy, or otherwise mutate a shared cluster.
Give the user a compact checklist of verified capabilities and remaining actions. Include the failing command category and error summary without secret values. Distinguish grants the user can request from local fixes. Do not edit IAM data or file a grant request unless the user asks.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Compare Zephyr changes using the same prepared dataset and matching execution settings. Analyze Finelog metrics for each processing stage to distinguish CPU efficiency, elapsed time, and peak memory changes, then report improvements, regressions, or tradeoffs.
Register Hugging Face datasets for Marin by inspecting their structure without downloading the full dataset. Identify text fields, configurations, and data splits, then add a dataset module and validate the mapping with a sample.
Turn a specified GCP access request into a reviewable Pulumi change and pull request. Select the narrowest suitable role and encrypt personal email identities, covering resource permissions and Cloud Run viewer access through IAP, an access gateway.
Add or update Pallas/Mosaic kernels, specialized computation routines for JAX. The skill covers reference implementations, output and gradient checks, public wrappers, fallback behavior, performance benchmarks, and hardware-specific tuning when requested.
Try model or training changes in a copied Grug or Grugformer experiment, document the differences, and bring successful changes back into the base template while removing superseded experiment code and checking behavior.
Organize Marin experiment reports into a maintained index, or archive explicitly named older experiments using lasting tags and issue references. Existing conclusions and canonical links are preserved so experiment records remain traceable.