policy-vla
Use when an approved task needs a VLA model to propose versioned action candidates from current visual and language context.
Skill by Chenxi-Li24
Chalk Labs
Need help deploying policy-vla?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/chenxi-li24-policy-vla
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: productivity
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/policy-vla
What's inside
- Skill name
policy-vla- Description
- Use when an approved task needs a VLA model to propose versioned action candidates from current visual and language context.
- License
- MIT
- Instructions
- 25 lines in SKILL.md · view on GitHub
Agent Skills spec issues
- name: "policy-vla" does not match folder "vla"
Files (2)
- SKILL.md
- manifest.yaml
SKILL.md (first lines)
# VLA Policy ## Purpose Produce candidate actions or action chunks for Task Engine review. The VLA is a policy provider, not the central task-planning LLM and not a hardware controller. ## Contract Input references the approved task, current `frameRef`, `robotStateRef`, targetRef and policy version. Output contains bounded candidate actions, confidence, model identity and observation revisions. Candidates are never sent directly to CAN. Task Engine binds them to a plan revision; Robot Service performs final limits and continuity checks. ## Safety A model change, stale observation, target revision change, missing Supervisor or candidate outside the authorized envelope invalidates execution. The Skill must support cancel and may not auto-recover a stopped task. ## Dependencies Requires Robot, Vision, Supervisor, Startouch, XVisio and a verified `policy.vla` model asset. ## Current Status No accepted worker or VLA weight is present. Registry must report both `implementation_not_migrated` and missing model; fake output cannot be marked ready.
From Chenxi-Li24/UIEAclub_ThirdHand_VLA, MIT.
More from Chenxi-Li24/UIEAclub_ThirdHand_VLA
- 0.565score
vision-supervise-execution
Chenxi-Li24
Use when an approved physical task needs independent visual checks for target visibility, alignment, grasp progress, or placement success.
SkillBFreshness: B★ 4Claude Code +9 - 0.565score
policy-act
Chenxi-Li24
Use when an approved task needs an ACT checkpoint to propose bounded, versioned action chunks from current observations.
SkillBFreshness: B★ 4Claude Code +9 - 0.565score
policy-diffusion-policy
Chenxi-Li24
Use when an approved task needs a diffusion-policy model to propose bounded action candidates from current observations.
SkillBFreshness: B★ 4Claude Code +9 - 0.565score
vision-describe-scene
Chenxi-Li24
Use when a task needs a structured, read-only summary of the current RGB-D scene before selecting an object or planning an action.
SkillBFreshness: B★ 4Claude Code +9 - 0.565score
vision-inspect-scene
Chenxi-Li24
Use when the ThirdHand controller needs a fresh, read-only natural-language observation from the existing XVisio RGB stream.
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
vision-detect-objects
Chenxi-Li24
Use when a task needs object classes, labels, confidence, and stable target references from the current RGB-D scene.
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
manipulation-pick-and-place
Chenxi-Li24
Use when the user requests a supervised pick of a stable target and placement at an approved destination.
SkillBFreshness: B★ 4Claude Code +9 - 0.304score
detect-bottles
Chenxi-Li24
No description from the publisher yet.
SkillBFreshness: B★ 4Claude Code +9 - 0.304score
select-target
Chenxi-Li24
No description from the publisher yet.
SkillBFreshness: B★ 4Claude Code +9 - 0.304score
gripper-control
Chenxi-Li24
No description from the publisher yet.
SkillBFreshness: B★ 4Claude Code +9
Signals
- Stars
- 4
- Forks
- 1
- Open issues
- 0
- Last commit (this folder)
- Sep 10, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from Chenxi-Li24
- 0.565score
vision-describe-scene
Chenxi-Li24
Use when a task needs a structured, read-only summary of the current RGB-D scene before selecting an object or planning an action.
SkillBFreshness: B★ 4Claude Code +9 - 0.565score
policy-act
Chenxi-Li24
Use when an approved task needs an ACT checkpoint to propose bounded, versioned action chunks from current observations.
SkillBFreshness: B★ 4Claude Code +9 - 0.565score
policy-diffusion-policy
Chenxi-Li24
Use when an approved task needs a diffusion-policy model to propose bounded action candidates from current observations.
SkillBFreshness: B★ 4Claude Code +9 - 0.565score
vision-inspect-scene
Chenxi-Li24
Use when the ThirdHand controller needs a fresh, read-only natural-language observation from the existing XVisio RGB stream.
SkillBFreshness: B★ 4Claude Code +9 - 0.565score
vision-supervise-execution
Chenxi-Li24
Use when an approved physical task needs independent visual checks for target visibility, alignment, grasp progress, or placement success.
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
manipulation-pick-and-place
Chenxi-Li24
Use when the user requests a supervised pick of a stable target and placement at an approved destination.
SkillBFreshness: B★ 4Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get policy-vla running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
policy-vla skill. Agent Mart. https://agentmart.dev/skills/chenxi-li24-policy-vla (last updated 2026-10-24)