Agusx1211

Generate and retain a deterministic Project New Shoes agent codename. Use at the start of repository tasks that may create, claim, comment on, review, or open a GitHub issue or pull request so same-model agents remain distinguishable.

Play a Command & Conquer Generals: Zero Hour match to its terminal outcome through the New Shoes authenticated direct-control API. Use for autonomous skirmish play when an agent needs a faction-neutral guide to game systems, semantic controls, capability discovery, event-driven cadence, camera-bound interaction, and authoritative outcome handling without a prescribed strategy or director scripts.

Set up, launch, record, supervise, and clean up a Project New Shoes Zero Hour agent match using the real WebAssembly runtime and authenticated AgentBridge. Use when preparing a local or remote hardware-GPU match, configuring Global or Camera direct-control mode, handing the REST session to a playing agent, preserving terminal evidence and video, or diagnosing orchestration and shutdown problems.

Decode, summarize, extract, and replay C&C Zero Hour browser issue dumps. Use when Codex receives or needs to inspect `.cncdump.json` / `.cncdump.json.zip` files from `harness/play.html`, diagnose a reported visual/input/gameplay bug from an issue dump, extract embedded screenshots/annotations/video/logs/deep snapshots, compare dump build metadata, or run the dump through `replay_issue_dump.mjs`.

Prepare and publish a Project New Shoes release by promoting the dev branch to main. Use when asked to release, cut a version, prepare a dev-to-main release PR, update the release changelog, or verify that a release PR completely accounts for its included pull requests.

Search the local reference-doc library in assets/docs (community ports, D3D8/W3D/format specs, browser API docs). Use when a question is about original-engine behavior, D3D8 semantics, file formats (BIG/W3D/WND/CSF/INI), Miles/Bink/GameSpy APIs, or WebGL2/Web Audio/WebCodecs/wasm guarantees — before reverse-engineering from scratch or searching the web.

Author a benchmark task (prompt + expected solution) for a pinned project. Use when the user says "create a task", "make a hard task on <project>", "add a scout/coder challenge", or asks to design an easy/medium/hard evaluation. Studies the repo, designs a difficulty-calibrated task, and writes task.md, solution.md, and task.json.

Run a coding agent blind against a benchmark task, review its work, score it on every metric, and write a report. Also runs stateful, resumable multi-task campaigns ("fully eval <model>") that tick a durable checklist case-by-case in serial order. Use when the user says "evaluate <model> on <task>", "fully eval <model>", "run the benchmark", "score an agent", "test <model> against <project>", "resume the eval". Spins up a throwaway git worktree in /tmp/work, dispatches the agent (pi-as-mcp by default), waits serially, then produces reports/<timestamp>-<model>/ with report.md, metrics.json, frozen request+solution, and diff.zip.

Import and pin a codebase into the benchmark. Use when the user says "import repo", "add a project to the bench", "pin <repo>", or gives a git URL/path to benchmark against. Clones the repo at a frozen commit, writes its SUMMARY.md orientation doc and project.json, and registers it in manifest.json.

Show the state of the benchmark — pinned projects, which of the six tasks each has, and how many evaluation reports exist per task. Use when the user says "list the bench", "what projects do we have", "show benchmark coverage", "what's been evaluated".

Create session-scoped, filesystem-backed artifacts for editable handoffs. Use whenever output is meant to be copied or reused and spans multiple lines, including messages, review comments, prompts, configuration, commands, and prose, or when the user asks for an editable artifact. Do not use for ordinary source edits in the current workspace or for concise answers that are easier to read inline.