Verify Rundale gameplay changes live by running Parish engine scripts and inspecting JSON outputs.
npx skills add https://github.com/dmooney/Rundale --skill parish-engine
Please help me install this Agent Skill. Skill: parish-engine Source: https://github.com/dmooney/Rundale/tree/main/.agents/skills/parish-engine Command: npx skills add https://github.com/dmooney/Rundale --skill parish-engine
It helps you verify that Rundale gameplay changes are actually live in the engine runtime, not merely passing tests.
--script mode and inspect structured JSON output for movement, look responses, and system commands.Run the default gameplay walkthrough to confirm a change is live by executing: parish-engine via the repo’s script harness on the walkthrough fixture.
High-intent search queries and answers about installing and using this skill.
You can verify gameplay changes by running fixture scripts in --script mode to inspect structured JSON outputs for movement and system responses, confirming changes are live in the engine runtime rather than just passing tests.
Detect regressions by running deterministic eval baselines and structural rubrics against JSON transcripts, checking for conditions like non-empty look descriptions and positive movement minutes to automatically identify gameplay regressions.
Yes, exploratory play-testing can be automated by autonomously exploring and stressing the game via script workflows, generating JSON transcripts evaluated against baselines to detect runtime issues and text quality regressions.
Yes, you must build the Parish engine with cargo build before executing script workflows. Building the Rust project first ensures the runtime environment is ready to process fixture scripts and generate structured JSON output.
Feature proving at runtime involves executing targeted scripts that exercise a single gameplay feature, then inspecting the JSON transcript to confirm world, NPC, and time changes alongside response text quality.
Passing tests does not guarantee changes are live in the engine runtime. Runtime verification addresses this by driving the engine through script workflows to validate actual outputs for movement and look rendering, ensuring changes are live.
gastownhallAutomate end-to-end smoke testing for the wasteland wl binary.
sunholo-dataAnalyze AILANG evaluation baselines to categorize failures and compare model performance.
Sir-chawakornBuilds an LLM evaluation harness using golden datasets and code-based scorers.
hatayamaSimulate keyboard input in Unity PlayMode using the Input System.