The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill clones the i4h-workflows repository from GitHub to establish the local development environment. The default source is a vendor-associated repository.
Evidence: git clone "$I4H_WORKFLOWS_REPO_URL" "$ROOT" in SKILL.md.
[COMMAND_EXECUTION]: The skill executes several shell commands and local scripts to manage the lifecycle of the simulation and authoring session.
Evidence: Execution of ./run.sh, ./local-agent/bridge.sh, and scripts/live_scene_edit.py to interact with the environment.
[DYNAMIC_EXECUTION]: The skill uses a Python bridge to send commands to a running Isaac Sim instance on port 8226 and generates code updates to 'bake' live changes into persistent source files.
Evidence: Use of isaacsim_send.py and the coding agent's logic to patch Scene and manifest templates based on live snapshots.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and modifies existing workflow files, including Python scripts and YAML manifests, based on external instructions.
Ingestion points: target workflow, Scene Python, scene manifest, embodiment manifest, and task manifests as specified in SKILL.md.
Boundary markers: No explicit prompt boundaries or delimiters are used for the content being processed.
Capability inventory: Includes repository cloning, local script execution, and multi-file source code modification.
Sanitization: Incorporates static validation using ruff and lint, as well as visual verification steps within the simulator before final persistence.

i4h-workflow-scene-edit