菜单
Set up Spell from a fresh checkout and run the first smoke tests or examples. Use when a user asks to install Spell, get started, verify the CLI, configure provider credentials, choose a first example, or try Spell for the first time.
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
Navigate and develop the Spell source code. Use when modifying Spell internals, investigating evaluator/runtime/provider behavior, finding the right source files, updating tests, or explaining how the implementation fits together.
Use Spell from the CLI, the Clojure API, and the Python benchmark adapter. Use when writing commands, embedding Spell in Clojure, calling Spell from benchmark Python code, or checking current API vocabulary.
Configure custom Spell agent and model profiles. Use when creating or modifying .agent.edn files, choosing namespaces, changing system prompts, wiring sub-agents, selecting provider-prefixed models, or doing agent smoke tests.
Navigate and develop the Spell source code. Use when modifying Spell internals, investigating evaluator/runtime/provider behavior, finding the right source files, updating tests, or explaining how the implementation fits together.
Set up Spell from a fresh checkout and verify prerequisites, CLI smoke tests, and available model provider authorization. Use when helping a user get to a runnable checkout, configure API provider access, or try the first Spell example.
Create, inspect, or modify Spell model profiles and agent profile files. Use when the user asks to configure providers, add or change `.agent.edn` files, expose namespaces, define sub-agents, adjust budgets, or use `spell.api/run`.
| name | spell-setup |
| description | Set up Spell from a fresh checkout and run the first smoke tests or examples. Use when a user asks to install Spell, get started, verify the CLI, configure provider credentials, choose a first example, or try Spell for the first time. |
Use this skill to make Spell usable from a fresh checkout. Prefer running commands directly and reporting the result.
Spell needs Java 11+, the Clojure CLI, and either a live provider credential or the test provider.
Check:
java -version
clj -Sdescribe
bin/spell -h
command -v spell || true
If Clojure is missing on macOS, recommend:
brew install clojure/tools/clojure
If spell does not already resolve to this checkout's bin/spell, ask the user
whether they want to add this checkout's bin/ directory to their shell PATH.
Do not edit shell startup files without asking. If they say yes, prefer appending
an absolute path entry to the appropriate shell profile and then verify with
command -v spell in a fresh shell.
For a no-cost smoke test, use the built-in test provider:
bin/spell -t "Return a short greeting"
Before asking the user which live provider to configure, check the current auth state without printing secret values:
test -n "${OPENAI_API_KEY:-}" && echo "OPENAI_API_KEY is set" || echo "OPENAI_API_KEY is not set"
test -n "${ANTHROPIC_API_KEY:-}" && echo "ANTHROPIC_API_KEY is set" || echo "ANTHROPIC_API_KEY is not set"
test -n "${FIREWORKS_API_KEY:-}" && echo "FIREWORKS_API_KEY is set" || echo "FIREWORKS_API_KEY is not set"
test -f "$HOME/.codex/auth.json" && echo "Codex auth exists" || echo "Codex auth not found"
Report which authorization options were found. If at least one is present, say which corresponding provider paths are ready to try and ask whether the user wants to configure any additional provider. If none are present, ask whether the user wants to configure one now and list these options:
codex once so ~/.codex/auth.json exists.OPENAI_API_KEY; use openai-tc:<model>.ANTHROPIC_API_KEY; use anthropic-pf:<model> or anthropic-tc:<model>.FIREWORKS_API_KEY; use fireworks:<model> or fireworks-tc:<model>.Model names must be exact. Do not guess model IDs; check provider docs before adding or recommending new model strings.
Start with examples that are small and easy to inspect:
bin/spell -e hello-world
bin/spell -v -e hello-world
bin/spell -e coin-flip
Use -b for a dollar budget and -d for maximum recursive LLM depth when running live examples:
bin/spell -b 0.25 -d 6 -e hello-world
Use -T to record a trace when the user wants to inspect execution:
bin/spell -T -v -e hello-world
README.md: human-facing overview and language introduction.AGENTS.md: source guide, setup commands, and reading order.examples/README.md: canonical public example list.examples/AGENTS.md: example-specific agent guidance.config/AGENTS.md: provider, model profile, and agent profile guide.