Skip to content

DexForce/EmbodiChain / .agents/skills/add-task-program

add-task-program

Verified

Create, compose, attach, validate, or repair an EmbodiChain Task Program. Use for program.yaml, integration.yaml, TaskProgramCfg, scene_binding, configured Task Program deployments, Semantic Call workflows, or Task Program expert demonstrations.

references/deployment.md

View on GitHub
# Task Program Deployment

Attach a program and integration to a physical environment, embodiment, and
execution policy through a thin runnable Gym config.

## Canonical selector

```yaml
id: MyTask-v1

environment:
  component: env.yaml

task_program:
  program: task_program/program.yaml
  integration: task_program/integration.yaml
  execution_policy: ../../../components/execution_policies/trajectory_open_loop.yaml

embodiment:
  component: ../../../components/embodiments/ur5_dh_pgi_140_80.yaml
```

All component paths resolve from the runnable deployment file. Use YAML
components. Do not add `task.component`; that compatibility boundary has been
removed.

For deployment-specific policy tuning, retain the shared base policy and use
the closed selector instead of duplicating its full configuration:

```yaml
execution_policy:
  component: ../../../components/execution_policies/motion_gen_verified.yaml
  overrides:
    motion:
      velocity_limit: 0.5
```

Only allowlisted motion/recovery scalar fields may be overridden; see the
task package README and the configured policy resolver. Contracts, planner,
tracking and effect assurance stay with the base policy. The original string
reference remains supported. Do not add policy inheritance or arbitrary merges.

## Composition rules

- Select either `environment.component` or corresponding inline environment
  and physical scene fields, never both.
- Select either `embodiment.component` or inline `robot`/`sensor`, never
  both.
- A configured Task Program must select an embodiment with `skill_profile`.
- An environment component is Task Program independent and may also serve a
  handwritten or RL deployment.
- A physical scene component contains no Task Program metadata.
- Embodiment robot installation `overrides` are limited to `uid`, `init_pos`,
  `init_rot`, and `init_qpos`. A separate UID-keyed `camera_overrides` may change
  only declared ordinary cameras' intrinsics/extrinsics; sensor membership,
  type, parent, resolution and outputs remain component-owned. See the
  [calibration contract](../../../../agent_context/topics/env-framework/configuration.md#reusable-gym-deployment-components).

## Multiple embodiments

Reuse one program across variants. Reuse one integration only when all
variants satisfy its required embodiment contract and task defaults address
logical resources shared by those profiles. Otherwise create an intentionally
different integration/policy pairing; do not fork the provider-independent
program just to rename robot parts.
A matching capability contract does not qualify the physical workspace, a fixed
object-to-EEF grasp, or a handover target for another tool. Check base placement,
open-hand approach, closing contacts and final physical acceptance before reusing
an integration. Keep those task choices separate from hardware TCP/contact-frame
calibration.

Give every runnable deployment a unique Gym ID. Loading a configuration-owned
ID is process-local and occurs when the config passes through
`config_to_cfg()`; task-package import alone does not register it.

## Python boundary

Do not create a task subclass for a fully configuration-defined deployment.
The loader registers the common `EmbodiedEnv`. If the task also provides a
separate handwritten or simulator-RL solution, its flat task module and
deployment can coexist under the same task identity.
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 319kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 118kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall