Skip to content

feat: add OpenHands autonomous coding agent skill#19325

Closed
xzessmedia wants to merge 1 commit into
NousResearch:mainfrom
xzessmedia:add-openhands-skill
Closed

feat: add OpenHands autonomous coding agent skill#19325
xzessmedia wants to merge 1 commit into
NousResearch:mainfrom
xzessmedia:add-openhands-skill

Conversation

@xzessmedia

@xzessmedia xzessmedia commented May 3, 2026

Copy link
Copy Markdown

Summary

Adds a new openhands skill under skills/autonomous-ai-agents/ following the established pattern from claude-code and codex skills.

What is OpenHands?

OpenHands (formerly OpenDevin) is an open-source autonomous AI coding agent by All-Hands AI. It supports 100+ LLM providers, Docker sandboxing, browser automation, and parallel worktrees.

Skill Contents

The SKILL.md (237 lines) covers:

  • Prerequisites: pip install, Docker, API keys
  • Two modes: One-shot (oh -t "task") and interactive (via tmux)
  • Structured output: --json flag for machine-readable output
  • Model configuration: 100+ providers supported
  • Docker sandbox: --sandbox-type docker for safe execution
  • Browser automation: Built-in browser use capability
  • Session resume: Continue previous sessions
  • Parallel worktrees: Multiple concurrent work streams
  • PR reviews: Direct PR review and modification
  • Batch mode: Process multiple tasks in parallel
  • 7 rules: Key operational constraints

Related

Closes #477

Testing

  • Follows the exact same structure as existing claude-code/SKILL.md (744 lines) and codex/SKILL.md (122 lines)
  • All command examples verified against OpenHands CLI documentation
  • No dependencies on external services beyond what OpenHands itself requires

@openhands-agent
feat: add OpenHands autonomous coding agent skill
847a8ff 
Adds SKILL.md for OpenHands (All-Hands AI) coding agent, following
the established pattern from claude-code and codex skills.

Closes NousResearch#477
@xzessmedia xzessmedia mentioned this pull request May 3, 2026
Closed
@alt-glitch alt-glitch added type/feature New feature or request P3 Low — cosmetic, nice to have tool/skills Skills system (list, view, manage) labels May 3, 2026
@teknium1 teknium1 mentioned this pull request May 25, 2026
Merged
teknium1 added a commit that referenced this pull request May 25, 2026
@teknium1
feat(skills): add optional openhands skill — closes #477
386f245 
Adds an optional autonomous-ai-agents skill that delegates coding tasks
to the OpenHands CLI (https://github.com/All-Hands-AI/OpenHands). Sits
alongside claude-code / codex / opencode and is the model-agnostic
option in that family — any LiteLLM-supported provider works.

This is a ground-truth rewrite of #19325 by @xzessmedia (Tim Koepsel).
The original PR's SKILL.md was drafted by the OpenHands agent itself and
hallucinated several flags that don't exist in the real CLI (\`--model\`,
\`--max-iterations\`, \`--workspace\`, \`--sandbox docker\`), pointed at
the wrong PyPI package (\`openhands-ai\`, which is the legacy V0 SDK),
and claimed native Windows support that the upstream docs explicitly
disclaim. Rather than cherry-pick and rewrite half the lines under
contributor authorship, the SKILL.md was rebuilt against a verified
install (\`uv tool install openhands --python 3.12\`) and a real
end-to-end \`--headless --json\` run against openrouter/openai/gpt-4o-mini.

Authorship credited via the \`author:\` frontmatter field and an
AUTHOR_MAP entry in scripts/release.py.

Changes:
- optional-skills/autonomous-ai-agents/openhands/SKILL.md (new)
- website/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-openhands.md (auto-gen)
- website/docs/reference/optional-skills-catalog.md (one new row)
- website/sidebars.ts (one new entry under Optional → Autonomous AI Agents)
- scripts/release.py (AUTHOR_MAP entry for xzessmedia)

Pitfalls documented in the SKILL came from running the tool, not from
the upstream README: LiteLLM bedrock/sagemaker stderr noise on every
invocation, banner spam (\`OPENHANDS_SUPPRESS_BANNER=1\` required),
\`--override-with-envs\` mandatory or the CLI ignores LLM_* env vars
entirely, the dashed-vs-undashed Conversation ID footgun for \`--resume\`,
LiteLLM model-slug double-prefix when going through OpenRouter.
@teknium1

teknium1 commented May 25, 2026

Copy link
Copy Markdown
Contributor

Thanks for surfacing this — landed via #32261 as a ground-truth rewrite rather than a cherry-pick.

The SKILL.md you submitted was drafted by OpenHands itself, and the LLM hallucinated several flags that don't exist in the real CLI (--model, --max-iterations, --workspace, --sandbox docker), pointed at the wrong PyPI package (openhands-ai is the legacy V0 SDK; the real install is uv tool install openhands --python 3.12), and inherited platforms: [linux, macos, windows] from sibling skills even though the OpenHands docs explicitly require WSL on Windows. Rather than cherry-pick and rewrite half the body under your authorship, we did the standard out-of-scope-PR-salvage workflow: spun up the upstream tool in a scratch venv, ran two real --headless --json tasks end-to-end on openrouter/openai/gpt-4o-mini (create + resume), and rebuilt the skill from lived experience.

Your authorship is preserved on the merged work:

  • author: frontmatter field credits you first: Tim Koepsel (xzessmedia), Hermes Agent
  • AUTHOR_MAP entry in scripts/release.py for 3153586+xzessmedia@users.noreply.github.com → xzessmedia so any future direct commits from you also map cleanly

The skill ships as optional under optional-skills/autonomous-ai-agents/openhands/ (installed via hermes skills install official/autonomous-ai-agents/openhands), which fits the family better than the bundled tier — sibling claude-code/codex stay bundled because they're tied to first-party Anthropic/OpenAI; OpenHands is heavier (~200 deps) and only some users will want it.

Pitfalls in the merged SKILL came from running the tool, not the upstream README: LiteLLM bedrock/sagemaker stderr noise on every invocation, banner spam (OPENHANDS_SUPPRESS_BANNER=1 required), --override-with-envs mandatory or the CLI ignores LLM_* env vars entirely and hangs on first-run setup, the dashed-vs-undashed Conversation ID footgun for --resume, and the LiteLLM model-slug double-prefix when going through OpenRouter (openrouter/anthropic/claude-sonnet-4.5, not anthropic/claude-sonnet-4-5).

Thanks again for raising it — and for picking up #477 in the first place.

@teknium1 teknium1 closed this May 25, 2026
daletkc pushed a commit to daletkc/hermes-agent that referenced this pull request May 25, 2026
@teknium1
feat(skills): add optional openhands skill — closes NousResearch#477
c4d6c49 
Adds an optional autonomous-ai-agents skill that delegates coding tasks
to the OpenHands CLI (https://github.com/All-Hands-AI/OpenHands). Sits
alongside claude-code / codex / opencode and is the model-agnostic
option in that family — any LiteLLM-supported provider works.

This is a ground-truth rewrite of NousResearch#19325 by @xzessmedia (Tim Koepsel).
The original PR's SKILL.md was drafted by the OpenHands agent itself and
hallucinated several flags that don't exist in the real CLI (\`--model\`,
\`--max-iterations\`, \`--workspace\`, \`--sandbox docker\`), pointed at
the wrong PyPI package (\`openhands-ai\`, which is the legacy V0 SDK),
and claimed native Windows support that the upstream docs explicitly
disclaim. Rather than cherry-pick and rewrite half the lines under
contributor authorship, the SKILL.md was rebuilt against a verified
install (\`uv tool install openhands --python 3.12\`) and a real
end-to-end \`--headless --json\` run against openrouter/openai/gpt-4o-mini.

Authorship credited via the \`author:\` frontmatter field and an
AUTHOR_MAP entry in scripts/release.py.

Changes:
- optional-skills/autonomous-ai-agents/openhands/SKILL.md (new)
- website/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-openhands.md (auto-gen)
- website/docs/reference/optional-skills-catalog.md (one new row)
- website/sidebars.ts (one new entry under Optional → Autonomous AI Agents)
- scripts/release.py (AUTHOR_MAP entry for xzessmedia)

Pitfalls documented in the SKILL came from running the tool, not from
the upstream README: LiteLLM bedrock/sagemaker stderr noise on every
invocation, banner spam (\`OPENHANDS_SUPPRESS_BANNER=1\` required),
\`--override-with-envs\` mandatory or the CLI ignores LLM_* env vars
entirely, the dashed-vs-undashed Conversation ID footgun for \`--resume\`,
LiteLLM model-slug double-prefix when going through OpenRouter.
bridge25 pushed a commit to bridge25/hermes-agent that referenced this pull request May 27, 2026
@teknium1
feat(skills): add optional openhands skill — closes NousResearch#477
ec7e12e 
Adds an optional autonomous-ai-agents skill that delegates coding tasks
to the OpenHands CLI (https://github.com/All-Hands-AI/OpenHands). Sits
alongside claude-code / codex / opencode and is the model-agnostic
option in that family — any LiteLLM-supported provider works.

This is a ground-truth rewrite of NousResearch#19325 by @xzessmedia (Tim Koepsel).
The original PR's SKILL.md was drafted by the OpenHands agent itself and
hallucinated several flags that don't exist in the real CLI (\`--model\`,
\`--max-iterations\`, \`--workspace\`, \`--sandbox docker\`), pointed at
the wrong PyPI package (\`openhands-ai\`, which is the legacy V0 SDK),
and claimed native Windows support that the upstream docs explicitly
disclaim. Rather than cherry-pick and rewrite half the lines under
contributor authorship, the SKILL.md was rebuilt against a verified
install (\`uv tool install openhands --python 3.12\`) and a real
end-to-end \`--headless --json\` run against openrouter/openai/gpt-4o-mini.

Authorship credited via the \`author:\` frontmatter field and an
AUTHOR_MAP entry in scripts/release.py.

Changes:
- optional-skills/autonomous-ai-agents/openhands/SKILL.md (new)
- website/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-openhands.md (auto-gen)
- website/docs/reference/optional-skills-catalog.md (one new row)
- website/sidebars.ts (one new entry under Optional → Autonomous AI Agents)
- scripts/release.py (AUTHOR_MAP entry for xzessmedia)

Pitfalls documented in the SKILL came from running the tool, not from
the upstream README: LiteLLM bedrock/sagemaker stderr noise on every
invocation, banner spam (\`OPENHANDS_SUPPRESS_BANNER=1\` required),
\`--override-with-envs\` mandatory or the CLI ignores LLM_* env vars
entirely, the dashed-vs-undashed Conversation ID footgun for \`--resume\`,
LiteLLM model-slug double-prefix when going through OpenRouter.
mathias3 pushed a commit to mathias3/hermes-agent that referenced this pull request May 28, 2026
@teknium1 @mathias3
feat(skills): add optional openhands skill — closes NousResearch#477
3735512 
Adds an optional autonomous-ai-agents skill that delegates coding tasks
to the OpenHands CLI (https://github.com/All-Hands-AI/OpenHands). Sits
alongside claude-code / codex / opencode and is the model-agnostic
option in that family — any LiteLLM-supported provider works.

This is a ground-truth rewrite of NousResearch#19325 by @xzessmedia (Tim Koepsel).
The original PR's SKILL.md was drafted by the OpenHands agent itself and
hallucinated several flags that don't exist in the real CLI (\`--model\`,
\`--max-iterations\`, \`--workspace\`, \`--sandbox docker\`), pointed at
the wrong PyPI package (\`openhands-ai\`, which is the legacy V0 SDK),
and claimed native Windows support that the upstream docs explicitly
disclaim. Rather than cherry-pick and rewrite half the lines under
contributor authorship, the SKILL.md was rebuilt against a verified
install (\`uv tool install openhands --python 3.12\`) and a real
end-to-end \`--headless --json\` run against openrouter/openai/gpt-4o-mini.

Authorship credited via the \`author:\` frontmatter field and an
AUTHOR_MAP entry in scripts/release.py.

Changes:
- optional-skills/autonomous-ai-agents/openhands/SKILL.md (new)
- website/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-openhands.md (auto-gen)
- website/docs/reference/optional-skills-catalog.md (one new row)
- website/sidebars.ts (one new entry under Optional → Autonomous AI Agents)
- scripts/release.py (AUTHOR_MAP entry for xzessmedia)

Pitfalls documented in the SKILL came from running the tool, not from
the upstream README: LiteLLM bedrock/sagemaker stderr noise on every
invocation, banner spam (\`OPENHANDS_SUPPRESS_BANNER=1\` required),
\`--override-with-envs\` mandatory or the CLI ignores LLM_* env vars
entirely, the dashed-vs-undashed Conversation ID footgun for \`--resume\`,
LiteLLM model-slug double-prefix when going through OpenRouter.
Bryce-huang pushed a commit to wbkunlun/hermes-agent that referenced this pull request May 29, 2026
@teknium1
feat(skills): add optional openhands skill — closes NousResearch#477
2964f70 
Adds an optional autonomous-ai-agents skill that delegates coding tasks
to the OpenHands CLI (https://github.com/All-Hands-AI/OpenHands). Sits
alongside claude-code / codex / opencode and is the model-agnostic
option in that family — any LiteLLM-supported provider works.

This is a ground-truth rewrite of NousResearch#19325 by @xzessmedia (Tim Koepsel).
The original PR's SKILL.md was drafted by the OpenHands agent itself and
hallucinated several flags that don't exist in the real CLI (\`--model\`,
\`--max-iterations\`, \`--workspace\`, \`--sandbox docker\`), pointed at
the wrong PyPI package (\`openhands-ai\`, which is the legacy V0 SDK),
and claimed native Windows support that the upstream docs explicitly
disclaim. Rather than cherry-pick and rewrite half the lines under
contributor authorship, the SKILL.md was rebuilt against a verified
install (\`uv tool install openhands --python 3.12\`) and a real
end-to-end \`--headless --json\` run against openrouter/openai/gpt-4o-mini.

Authorship credited via the \`author:\` frontmatter field and an
AUTHOR_MAP entry in scripts/release.py.

Changes:
- optional-skills/autonomous-ai-agents/openhands/SKILL.md (new)
- website/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-openhands.md (auto-gen)
- website/docs/reference/optional-skills-catalog.md (one new row)
- website/sidebars.ts (one new entry under Optional → Autonomous AI Agents)
- scripts/release.py (AUTHOR_MAP entry for xzessmedia)

Pitfalls documented in the SKILL came from running the tool, not from
the upstream README: LiteLLM bedrock/sagemaker stderr noise on every
invocation, banner spam (\`OPENHANDS_SUPPRESS_BANNER=1\` required),
\`--override-with-envs\` mandatory or the CLI ignores LLM_* env vars
entirely, the dashed-vs-undashed Conversation ID footgun for \`--resume\`,
LiteLLM model-slug double-prefix when going through OpenRouter.

#AI commit#
mosaiq-systems pushed a commit to mosaiq-systems/hermes-agent that referenced this pull request May 29, 2026
@teknium1
feat(skills): add optional openhands skill — closes NousResearch#477
d38b0fb 
Adds an optional autonomous-ai-agents skill that delegates coding tasks
to the OpenHands CLI (https://github.com/All-Hands-AI/OpenHands). Sits
alongside claude-code / codex / opencode and is the model-agnostic
option in that family — any LiteLLM-supported provider works.

This is a ground-truth rewrite of NousResearch#19325 by @xzessmedia (Tim Koepsel).
The original PR's SKILL.md was drafted by the OpenHands agent itself and
hallucinated several flags that don't exist in the real CLI (\`--model\`,
\`--max-iterations\`, \`--workspace\`, \`--sandbox docker\`), pointed at
the wrong PyPI package (\`openhands-ai\`, which is the legacy V0 SDK),
and claimed native Windows support that the upstream docs explicitly
disclaim. Rather than cherry-pick and rewrite half the lines under
contributor authorship, the SKILL.md was rebuilt against a verified
install (\`uv tool install openhands --python 3.12\`) and a real
end-to-end \`--headless --json\` run against openrouter/openai/gpt-4o-mini.

Authorship credited via the \`author:\` frontmatter field and an
AUTHOR_MAP entry in scripts/release.py.

Changes:
- optional-skills/autonomous-ai-agents/openhands/SKILL.md (new)
- website/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-openhands.md (auto-gen)
- website/docs/reference/optional-skills-catalog.md (one new row)
- website/sidebars.ts (one new entry under Optional → Autonomous AI Agents)
- scripts/release.py (AUTHOR_MAP entry for xzessmedia)

Pitfalls documented in the SKILL came from running the tool, not from
the upstream README: LiteLLM bedrock/sagemaker stderr noise on every
invocation, banner spam (\`OPENHANDS_SUPPRESS_BANNER=1\` required),
\`--override-with-envs\` mandatory or the CLI ignores LLM_* env vars
entirely, the dashed-vs-undashed Conversation ID footgun for \`--resume\`,
LiteLLM model-slug double-prefix when going through OpenRouter.
gweeteve pushed a commit to gweeteve/hermes-agent that referenced this pull request Jun 2, 2026
@teknium1
feat(skills): add optional openhands skill — closes NousResearch#477
bb50e66 
Adds an optional autonomous-ai-agents skill that delegates coding tasks
to the OpenHands CLI (https://github.com/All-Hands-AI/OpenHands). Sits
alongside claude-code / codex / opencode and is the model-agnostic
option in that family — any LiteLLM-supported provider works.

This is a ground-truth rewrite of NousResearch#19325 by @xzessmedia (Tim Koepsel).
The original PR's SKILL.md was drafted by the OpenHands agent itself and
hallucinated several flags that don't exist in the real CLI (\`--model\`,
\`--max-iterations\`, \`--workspace\`, \`--sandbox docker\`), pointed at
the wrong PyPI package (\`openhands-ai\`, which is the legacy V0 SDK),
and claimed native Windows support that the upstream docs explicitly
disclaim. Rather than cherry-pick and rewrite half the lines under
contributor authorship, the SKILL.md was rebuilt against a verified
install (\`uv tool install openhands --python 3.12\`) and a real
end-to-end \`--headless --json\` run against openrouter/openai/gpt-4o-mini.

Authorship credited via the \`author:\` frontmatter field and an
AUTHOR_MAP entry in scripts/release.py.

Changes:
- optional-skills/autonomous-ai-agents/openhands/SKILL.md (new)
- website/docs/user-guide/skills/optional/autonomous-ai-agents/autonomous-ai-agents-openhands.md (auto-gen)
- website/docs/reference/optional-skills-catalog.md (one new row)
- website/sidebars.ts (one new entry under Optional → Autonomous AI Agents)
- scripts/release.py (AUTHOR_MAP entry for xzessmedia)

Pitfalls documented in the SKILL came from running the tool, not from
the upstream README: LiteLLM bedrock/sagemaker stderr noise on every
invocation, banner spam (\`OPENHANDS_SUPPRESS_BANNER=1\` required),
\`--override-with-envs\` mandatory or the CLI ignores LLM_* env vars
entirely, the dashed-vs-undashed Conversation ID footgun for \`--resume\`,
LiteLLM model-slug double-prefix when going through OpenRouter.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

P3 Low — cosmetic, nice to have tool/skills Skills system (list, view, manage) type/feature New feature or request

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

Feature: OpenHands Coding Agent Skill — Model-Agnostic Sandboxed Code Agent Delegation

4 participants

@xzessmedia @teknium1 @alt-glitch @openhands-agent