Add MAPPO to multi-agent documentation commands

[kellyguo11]

Description

The environment browser advertises SKRL support for the native multi-agent tasks but previously omitted the algorithm from its generated commands. SKRL consequently used its PPO default, converted the environment to single-agent form, and selected the wrong algorithm for pretrained-checkpoint lookup.

This change marks MAPPO as the generated-command default whenever a task registers an SKRL MAPPO configuration, then appends --algorithm MAPPO in the environment browser. Explicit runtime algorithm selection remains unchanged, so other algorithm checkpoints are still supported.

This currently updates the generated commands for:

• Isaac-Pendulum-MARL-Direct
• Isaac-Shadow-Handover-Direct

Tracks NVBug 6675391. No matching existing PR was found; #7360 addresses the separate benchmark play environment-creation path.

No new dependencies are required.

Type of change

• Bug fix (non-breaking change which fixes an issue)
• Documentation update

Release backport

• Backport this pull request to the active release branch after it merges into develop

Screenshots

Not applicable. A headless browser smoke test generated:

uv run --extra skrl isaaclab train --rl_library skrl --task Isaac-Shadow-Handover-Direct --algorithm MAPPO physics=newton_mjwarp

Validation

• uv run --extra test python -m pytest --confcutdir=tools/test tools/test/test_environ_docs.py -q (28 passed)
• uv run --isolated --extra dev --extra ov -- make -C docs current-docs (passed)
• uv run isaaclab -f (passed)
• Headless Chrome environment-browser command generation smoke test (passed)

Checklist

• I have read and understood the contribution guidelines
• I have run the repository formatting and pre-commit checks with uv run isaaclab -f
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove the fix is effective
• No changelog fragment is required because no source package changed
• My name already exists in CONTRIBUTORS.md

[greptile-apps]

Greptile Summary

This PR updates generated environment-browser commands to select SKRL MAPPO for tasks that register a MAPPO configuration.

• Derives per-library default algorithm metadata while generating browser task rows.
• Adds --algorithm MAPPO when SKRL is selected for the two affected multi-agent tasks.
• Adds focused generator coverage for MAPPO metadata and refreshes the generated browser data.

Confidence Score: 5/5

The PR appears safe to merge; the generated MAPPO argument is correctly scoped to compatible SKRL tasks and accepted by both command modes.

No actionable failures remain: positional row generation is consistent, command state does not retain stale algorithms, and MAPPO resolves to registered configurations for both affected tasks.

Important Files Changed

Filename	Overview
docs/source/_static/css/environment-browser.js	Extends task metadata with default algorithms and appends the applicable algorithm to generated commands.
tools/environ_docs.py	Detects registered SKRL MAPPO configurations and emits correctly padded default-algorithm metadata.
tools/test/test_environ_docs.py	Verifies that tasks offering MAPPO render it as the generated SKRL command default.

_{Reviews (1): Last reviewed commit: "Add MAPPO to multi-agent documentation c..." | Re-trigger Greptile}

[isaaclab-review-bot]

Isaac Lab Review Bot

Reviewed the environment-browser row-schema extension that records MAPPO as SKRL’s generated-command default for registered multi-agent tasks. The generated rows and command builder consistently produce --algorithm MAPPO for the two affected tasks without changing other generated commands.

• Design and architecture: The generator derives an optional per-library default-algorithm map from registered SKRL configurations, while the browser remains responsible for command assembly. This preserves the existing generated-data/consumer split. MAPPO preference is currently encoded as generator policy, so supporting a task that registers MAPPO but prefers another default would require refining that policy.
• API: The positional row format remains compatible with existing rows: JavaScript defaults the new ninth field to an empty object, prior preview and warp slots retain their positions, and preview-image collection continues reading index 6. Explicit runtime algorithm selection is unaffected.
• Implementation: The revised padding conditions correctly emit the optional compatibility, preview, warp, and algorithm fields in sequence. The regenerated task rows match the new renderer, and command generation adds the algorithm only when the selected RL library has a configured default. The focused regression test covers MAPPO metadata emission.

No blocking issues. No inline issue met the actionable-evidence threshold; the assessment above records the review feedback.

Automated review; human maintainers own approval decisions.

[kellyguo11]

run-ci