Skill development, execution reliability and the Opper gateway

Changes since v1.1.0

Added

Opper as an OpenAI-compatible gateway, with provider selection, environment
credentials and model identifiers preserved in routed requests.
Bilingual Track B development guides and an independent Piper X simulation
Skill example covering model conversion, original Node locks, packaging,
and validation with its recorded runtime baseline.
Bilingual contribution guides, MuJoCo Quickstart GIFs, and a GOSIM Shenzhen
2026 announcement.

Local workflow-document and Node-executable modifications produce warnings
during task creation and Runtime startup. Bindings freeze the activated
workflow document hash and experience records use it for attribution.
Consolidate bilingual manuals, developer references and documentation
navigation while retaining benchmark documentation links.
Synchronize package, CLI, bundled bridge, Docker, release-build defaults,
README and current documentation versions to 1.1.1.

Keep explicit Node verification and install-cache acceptance strict for
modified executables, and retain Skill/Runtime version compatibility checks.
Record structured Gateway rejections known to occur before Action dispatch
as failed, allowing a later retry without the unknown-execution guard.
Timeouts, transport failures and identified remote invocations remain unknown.
Exclude pre-dispatch rejections from task finalization and cancellation
execution facts; a task with no executed calls still fails. Structured
admission rejection codes require Forge Gateway 1.1.2 or newer; older
unstructured responses retain conservative handling.
Add regression coverage for Opper routing, local Node modifications and
pre-dispatch Action rejection handling.

Built and verified by GitHub Actions in an Ubuntu 20.04 container:

Python 3.11 and 3.12: 488 tests passed on each version, plus Ruff checks and bytecode compilation
Node 20 bridge build
Package metadata validation with twine check
Clean-environment wheel installation
paos --version and paos --help smoke tests

Install the attached wheel with:

pip install ./phyagentos_ai-1.1.1-py3-none-any.whl