setup-pstack
Configure which models pstack uses per role. Writes a config file that overrides the skill defaults, and imports it into your global CLAUDE.md so every session sees it. Use for /setup-pstack, "configure pstack models", or changing pstack's model choices.
- License 许可
- MIT
- 最近更新
- 2026-08-20
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- setup-pstack
- description
- Configure which models pstack uses per role. Writes a config file that overrides the skill defaults, and imports it into your global CLAUDE.md so every session sees it. Use for /setup-pstack, "configure pstack models", or changing pstack's model choices.
03 SKILL.md 正文
Setup pstack
Write ~/.claude/pstack-models.md, a per-role model map, and import it from ~/.claude/CLAUDE.md so it is in context for every session. The skills read it and fall back to their inline defaults when a line is absent, so this is an override layer, not a requirement.
Steps
1. Know the roster
Claude Code's Agent tool takes one of four models: opus, sonnet, haiku, fable. That is the whole set. Do not write any other value.
Two aliases are also valid: inherit-parent and auto. Both mean the role runs on the parent chat model, which is how a user on a single model stays on it. They are not model names; they tell the skill to omit the Agent model param.
Effort is a separate axis. An Agent call can carry effort: low | medium | high, and a role line may name one after the model, as opus/high. A role with no effort suffix uses the default.
2. Load current state
The default role-to-model mapping is the file shape shown in step 4. If ~/.claude/pstack-models.md already exists, read it and treat its values as the current choices. Otherwise start from those defaults.
3. Map and confirm
Show every role with its current model. Ask whether to accept as-is or change specific roles, offering opus, sonnet, haiku, fable, inherit-parent, and auto. Prefer AskUserQuestion over free text.
For panel roles (how critics, arena runners, architect runners, interrogate reviewers) the value is a list, and one subagent runs per entry, alias entries included, so the list length sets the count. arena cross-judge pool is also a list, and Arena picks one entry from it, preferring one that differs from the parent's model. swarm workers is the default model for every worker unless a race or comparison assigns another model per arm.
Repeated entries in a panel are allowed and sometimes right. Four arms on the same model still diverge, because sampling does. But a panel whose whole point is contrast reads better with four distinct models.
4. Write the config
Write ~/.claude/pstack-models.md. Overwrite the whole file so re-runs stay idempotent. Shape:
# pstack model configuration. One line per role. Delete a line to fall back to the skill default.
# Valid values: opus, sonnet, haiku, fable. Add an effort suffix as `opus/high` when the role wants it.
# `inherit-parent` or `auto` as a value: the role runs on the parent chat model (omit the `Agent` `model` param).
# Alias entries in a panel list still count toward its fan-out.
feature, refactoring: sonnet
bug-fix: opus/high
perf-issue: opus/high
hillclimb: opus/high
judgment and prose: fable
hardest tasks: fable
how explorer: sonnet
how explainer: fable
how critics: fable, opus, sonnet, haiku
why investigators: sonnet
why synthesizer: fable
reflect tooling: opus
reflect judgment, divergent, synthesizer: fable
arena runners: fable, opus, sonnet, haiku
arena cross-judge pool: fable, opus, sonnet, haiku
swarm workers: sonnet
architect runners: fable, opus, sonnet, haiku
interrogate reviewers: fable, opus, sonnet, haiku
5. Import it into CLAUDE.md
Claude Code has no always-applied rule file, so the config has to be imported or every skill pays a Read to find it.
Read ~/.claude/CLAUDE.md. If it already contains a line @pstack-models.md, leave it alone. Otherwise append that line on its own, under a # pstack heading if the file has headings. Never write the line twice, and never rewrite anything else in that file. If ~/.claude/CLAUDE.md does not exist, create it containing just the heading and the import.
A user who declines the import keeps the config file. Tell them the skills will still find it, at the cost of one Read per invocation, and that they can add @pstack-models.md themselves later.
6. Confirm
Tell the user the config was written and that the import applies to new sessions. Re-running this skill updates both.
7. Offer a verification skill (optional)
Check whether the project has a way to drive the real app for proof (a verify-* skill, or an existing harness). If not, offer once: "want a project-local verification skill, so agents can drive the app the way a user does and prove changes work? I can generate one with /create-verification-skill." On yes, invoke /create-verification-skill (resolves wherever pstack is installed, whether project, user, or plugin). On no, move on without pushing.
04 同仓库更多 Skills
why
Use for 'why does X work this way', 'why we picked Y', design rationale, regressions, postmortems, or data-backed thresholds. Discovers available MCPs and queries each evidence category (source control, issue tracker, long-form docs, real-time chat, infrastructure observability, error tracking, product analytics warehouse) in parallel, then returns a cited read on decisions and tradeoffs. Use how for runtime behavior.
unslop
Cut AI tells from any writing. Must always apply.
typescript-best-practices
TypeScript best practices. Use when reading or editing any .ts or .tsx file.
technical-writing
Layered technical-writing standard: Diátaxis structure, Google developer style sentences, STE instruction rules, Global English syntax. Use for /technical-writing or when writing or reviewing docs, RFCs, readmes, PR descriptions, or commit messages.
teach
Explain a body of work plainly so a person actually understands it. Runs the `how` and `why` skills and weaves what they find into one clear explanation. Use for 'teach me this', 'help me really understand X', 'explain this change or subsystem to me'.
tdd
Use only when the user explicitly asks for TDD, a failing test, or a regression test, OR when the bug has an obvious cheap local test target. Skip when the test path is unclear, expensive, integration-heavy, or not requested.