docs: add an Environments section to the user guide - #1833
Merged
Conversation
Contributor
|
Caution The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased. |
This was referenced Jul 28, 2026
Give environment-based training a home in the docs, separate from the core-concepts track. The overview page maps the three ways an environment reaches Miles (dataset + reward, self-wired code, external ecosystems), lists the existing connectors (Harbor, OpenEnv, Strands, tau-bench) and sandbox providers (Daytona), and encodes the three rollout plug-in layers in a per-capability ownership matrix. Harbor and OpenEnv get short pages pointing at their experimental recipes as the authoritative walkthroughs. Further providers slot in as table rows and pages of this group. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
nblintao
force-pushed
the
tao/docs-environments-section
branch
from
July 28, 2026 14:46
7809bbd to
c4aa984
Compare
Shi-Dong
approved these changes
Jul 28, 2026
Shi-Dong
left a comment
Shi-Dong
left a comment
Contributor
There was a problem hiding this comment.
LGTM with minor comments.
Comment on lines
+29
to
+31
| | Sandbox provider | Used within | | ||
| |---|---| | ||
| | [Daytona](https://www.daytona.io/) | OpenEnv, Harbor | |
Contributor
There was a problem hiding this comment.
I would replace this table with text
Users can choose their own sandbox provide, for example, Daytona, ...
Having a table here with a single row "Daytona" might give people the false impression that we have a special relationship with them lol.
Contributor
Author
There was a problem hiding this comment.
This is a placeholder. I'll very soon add more items here like agentENV and modal
Comment on lines
+21
to
+25
| The maintained recipe lives in | ||
| [`examples/experimental/swe-agent-v2`](https://github.com/radixark/miles/tree/main/examples/experimental/swe-agent-v2), | ||
| with synchronous and fully-async launchers. Follow the | ||
| [recipe README](https://github.com/radixark/miles/blob/main/examples/experimental/swe-agent-v2/README.md) | ||
| for the architecture, Harbor server setup, task format, and launch scripts. |
Contributor
There was a problem hiding this comment.
This example is stale and I have a new one introduced in PR 1741. But let's merge your PR first, and I'll update this section in my PR.
nblintao
added a commit
to willccbb/miles
that referenced
this pull request
Jul 30, 2026
Apply the review feedback from radixark#1739: retitle the page to 'Verifiers (Prime Intellect)' to disambiguate from RL reward verifiers, relocate it from the core User Guide track into the Environments group (added in radixark#1833) with a connectors-table row, drop the user-guide index row (second-level pages are not listed there), and rename the cli-reference section to 'Environment Integrations' so flags of other environment integrations can share it. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
nblintao
added a commit
to willccbb/miles
that referenced
this pull request
Jul 31, 2026
… compatibility - Retitle the docs page to 'Verifiers (Prime Intellect)' and move it into the user guide's Environments section (radixark#1833), with a connectors-table row; rename the cli-reference section to 'Environment Integrations'. - Reject --verifiers-config combined with a custom --rollout-function-path or with multi-LoRA instead of silently overriding them, with tests. - Survive a CI image without the optional [verifiers] extra: importorskip guards for the tests that import renderers or verifiers at call time, and read verifiers_config with getattr for the partial-Namespace callers in main's validation tests. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Give environment-based training a provider-neutral home in the docs, as a new Environments group in the User Guide tab, separate from the core-concepts track:
user-guide/environments(overview): the three ways an environment reaches Miles (a dataset plus a reward / your own code on the plug points / an external ecosystem), a table of the existing connectors (Harbor, OpenEnv, Strands Agents, τ-bench), the sandbox-provider axis (Daytona, used within OpenEnv and Harbor), and a per-capability ✓/○ ownership matrix over the three nested rollout plug-in layers (agent / generate / rollout function).user-guide/harboranduser-guide/openenv: short pages — what the framework is, how the integration works, and where the maintained experimental recipe lives. Details stay in the recipe READMEs so they are maintained in one place.Future external integrations slot in as a connectors-table row plus a page of this group (e.g. #1739 lands as a rollout-function entry), and future sandbox providers (E2B, ...) as a row of the sandbox table.
Validation
docs.jsonvalidated as JSON; navigation follows the existing group conventions (root+pages).openenv_launch_common.py/swe-agent-v2/run.py, reward paths (--custom-rm-pathin both recipes), and the recipe READMEs.🤖 Generated with Claude Code