菜单
Use when building, debugging, or reviewing CLI commands across nuggetizer, ragnarok, rank_llm, or umbrela and you need the shared castorini.cli.v1 envelope, common introspection commands, artifact shapes, or cross-repo CLI consistency rules.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Install and run Anserini quickly by downloading the latest fatjar from the official Maven Central repo instead of building from source. Use when users want fast setup, smoke tests, or command execution without Maven project compilation.
Use when onboarding to nuggetizer, ragnarok, rank_llm, or umbrela and you need development environment setup for one repo or several repos at once, including clone-if-needed, uv or pip installation paths, shared virtualenv reuse, and smoke tests.
Use when serving Anserini retrieval together with any subset of rank_llm, ragnarok, nuggetizer, or umbrela over HTTP, especially for local port planning, direct request payload compatibility, curl or jq pipelines, or sequencing retrieval, reranking, generation, nugget creation, nugget assignment, and passage judging from an Anserini fatjar RestServer.
Use when coordinating an end-to-end Castorini pipeline across rank_llm, ragnarok, nuggetizer, and umbrela, especially for stage handoffs, JSONL compatibility, retrieval-to-answer evaluation flow, or reproducing a multi-stage experiment.
Use when publishing nuggetizer, ragnarok, rank_llm, or umbrela to PyPI or TestPyPI and you need the release sequence for version bumps, build checks, twine validation, TestPyPI dry-runs, or final production publishing.
| name | castorini-cli-reference |
| description | Use when building, debugging, or reviewing CLI commands across nuggetizer, ragnarok, rank_llm, or umbrela and you need the shared castorini.cli.v1 envelope, common introspection commands, artifact shapes, or cross-repo CLI consistency rules. |
| metadata | {"version":"0.1.0","visibility":"public"} |
Quick reference for the shared CLI contract across the four Python Castorini repos: nuggetizer, ragnarok, rank_llm, and umbrela.
All four repos expose a repo-named binary (nuggetizer, ragnarok, rank-llm, umbrela) with converged introspection commands and the castorini.cli.v1 JSON envelope.
The flag matrix is closest across nuggetizer, ragnarok, and umbrela. rank_llm shares the introspection surface and envelope shape, but some execution flags and artifact naming are repo-specific.
When a task depends on the current contract shape, start with <repo> describe ..., <repo> schema ..., or <repo> doctor --output json before assuming a sibling repo matches remembered behavior.
Every repo exposes:
| Command | Purpose |
|---|---|
describe <cmd> | Machine-readable contract: flags, defaults, enums, examples |
schema <name> | JSON Schema for inputs, outputs, envelope |
doctor | Environment and dependency preflight |
validate <target> | Input/artifact validation without model execution |
prompt list|show|render | Inspect and render prompt templates |
view <path> | Inspect existing artifact files |
Plus repo-specific execution verbs:
create, assign, metricsgenerate, validate, convertrerank, evaluate, analyze, retrieve-cache, servejudge, evaluateRead these on demand for detailed specifications:
references/envelope-schema.md — Full castorini.cli.v1 envelope specificationreferences/shared-flags.md — Common flags and introspection commands across the shared Castorini CLIsreferences/artifact-shape.md — Artifact metadata shape and conventions--input-file, not --input_file). Legacy compatibility scripts may still use snake_case internally.--output vs --output-file: --output controls format (text|json|jsonl); --output-file controls destination path. These are independent.--resume, --overwrite, and --fail-if-exists cannot be combined.--dry-run vs --validate-only: --dry-run resolves inputs and reports what would happen. --validate-only checks the declared contract (schemas, types) without resolving resources.errors array in the JSON envelope.rank_llm command names differ: its primary execution verbs are rerank and evaluate, not generate or judge.<repo> schema <name> or <repo> describe <command>.