feat(cli): improve provider discoverability for agents and developers
Nobody has claimed this yet.
- Dominant language
- Rust
- Stars
- 8.7k
- Forks
- 1.3k
- Avg merge
- 2d 11h
- Merged PRs (30d)
- 253
Description
Problem Statement
AI agents building OpenShell workflows cannot discover the provider credential-hiding path. The provider system is powerful but invisible — agents take the path of least resistance (--env) because --provider requires multi-step setup knowledge that isn't in their context.
Human developers face the same problem: the CLI help for sandbox create doesn't prioritize --provider over --env for credentials, error messages don't suggest providers when connections fail, and there's no command to bridge from "I have these env vars" to "here are the providers I should use."
Proposed Design
1. openshell provider suggest command
New subcommand that matches env var names to built-in profile credential declarations:
$ openshell provider suggest --env JIRA_API_TOKEN=... --env MLFLOW_TRACKING_TOKEN=...
Suggested providers:
JIRA_API_TOKEN → jira (built-in profile)
MLFLOW_TRACKING_TOKEN → mlflow (built-in profile)
To set up:
openshell provider create --name my-jira --type jira --credential JIRA_API_TOKEN
openshell provider create --name my-mlflow --type mlflow --credential MLFLOW_TRACKING_TOKEN
This gives agents a direct command to discover the right path without reading docs.
2. Updated CLI help text
Reorder openshell sandbox create --help to recommend --provider before --env for credentials. Currently --env appears without context about its security implications.
3. Connection failure hints
When a sandbox connection to a host fails and that host matches a built-in profile's endpoint, include a hint in the error message suggesting the matching profile.
Dependencies
- Depends on NVIDIA/OpenShell#2547 (built-in Jira/MLflow profiles) —
provider suggestneeds profiles to match against - Better experience with NVIDIA/OpenShell#2548 (CLI credential warnings) — complementary discoverability mechanisms
- Better experience with NVIDIA/OpenShell#2545 (providers v2 default-on) — suggested provider path works without extra setup
Motivation
In opendatahub-io/agent-ops#7, the AI agent building the demo workflow did not know about providers v2. It passed all credentials as --env because that path is obvious and immediately works. provider suggest would have bridged the gap.
Agent Investigation
- Built-in profile credential env var names available via
builtin_profiles()incrates/openshell-providers/src/profiles.rs openshell provider list-profilesalready exists —suggestwould add env-var-to-profile matching- CLI help templates in
crates/openshell-cli/src/main.rs(PROVIDER_EXAMPLES, sandbox create help)
Checklist
- Searched for existing issues — no issue tracks provider discoverability for agents
- Design includes three complementary discoverability mechanisms
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with builtin_profiles() in crates/openshell-providers/src/profiles.rs and the existing provider list-profiles command, then read the CLI help templates in crates/openshell-cli/src/main.rs. The work is complete when provider suggest matches environment variable names to built-in profiles, sandbox create help prioritizes --provider, and matching connection failures include a profile hint.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust
- Domain
- cli, security
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100