climaker
暂无简介
- License 许可
- MIT
- 最近更新
- 2026-11-06
- Stars
- ★ 4
- Forks
- 0
02 资源文件 3
- 参考文档references/augmentation-registry.md4.8 KB
- 脚本tools/har2spec.sh3.1 KB
- 参考文档tools/README.md6.0 KB
03 SKILL.md 正文
climaker — CLI Printing Press wrapper
/climaker is a memorable entry point to the CLI Printing Press (mvanhorn/cli-printing-press): a Go generator + a set of Claude Code skills that read an API's docs (or sniff a website with no docs), absorb every competing CLI/MCP feature, and print a token-efficient Go CLI plus an MCP server with a local-SQLite data layer, FTS5 search, compound insight commands, and agent-native flags.
This skill does not reimplement the press. The press ships its own well-built /printing-press* skills. /climaker (1) installs the tool if missing, then (2) routes your request to the correct underlying skill or binary subcommand. Know what not to build — wrap the real tool.
Quick Start
/climaker Notion # print a CLI+MCP for an API by name
/climaker https://postman.com/explore # ...or point at a website (no spec needed)
/climaker --har ./capture.har # ...or a HAR export from DevTools
If the press isn't installed yet, run the Preflight below first, then re-issue the command — it expands to the /printing-press skill the press provides.
Preflight — ensure the press is installed
Two parts are required: the binary (cli-printing-press) and the Printing Press skills (/printing-press …). Check both; install whatever is missing.
# 1. Binary present?
cli-printing-press --version 2>/dev/null || echo "BINARY MISSING"
# 2. Skills present? (look for the /printing-press slash command)
npx -y skills@latest list -g -a claude-code 2>/dev/null | grep -qi printing-press \
&& echo "SKILLS OK" || echo "SKILLS MISSING"
Install (does both — binary via go install, then refreshes all press skills):
curl -fsSL https://raw.githubusercontent.com/mvanhorn/cli-printing-press/main/scripts/install.sh | bash
One side only:
curl -fsSL https://raw.githubusercontent.com/mvanhorn/cli-printing-press/main/scripts/install.sh | bash -s -- --cli-only
curl -fsSL https://raw.githubusercontent.com/mvanhorn/cli-printing-press/main/scripts/install.sh | bash -s -- --skills-only
Codex target instead of Claude Code: append --agent codex.
Prerequisites: Go 1.26.4+, Node/npm (for npx), Claude Code. Verify with cli-printing-press --version. If command not found after a good go install, add $GOPATH/bin (default ~/go/bin) to your PATH.
After a skills install/refresh you MUST reload the agent session so the new /printing-press slash commands load. Tell the user to restart Claude Code, then re-run their /climaker … request.
Routing — map the request to the press
After Preflight passes, translate the user's intent to the underlying command and invoke it. The press's slash commands ARE the supported interface; /climaker is the alias that gets you there.
| User intent | Run |
|---|---|
| Make / print a CLI for an API or website | /printing-press <name-or-URL> |
| ...with 60% fewer Opus tokens (offload codegen to Codex) | /printing-press <name> codex |
| ...from a HAR capture (site uses websockets/bot-detection) | /printing-press --har ./capture.har |
| Reprint an existing CLI under the latest machine | /printing-press-reprint <name> |
| Targeted fix-up of a generated CLI (diagnostics → fixes → clean) | /printing-press-polish <name> |
| Publish a finished CLI to the public library (validate → PR) | /printing-press-publish <name> |
| Turn friction from a dogfooding session into a library PR | /printing-press-amend [name] |
| Score / dogfood / verify a CLI dir (any CLI, even hand-built) | binary subcommands below |
| Check which env tokens a printed CLI needs | cli-printing-press auth doctor |
The press also ships printing-press-catalog, printing-press-import, printing-press-output-review, printing-press-retro, and printing-press-score skills — surface those if the request matches (browse catalog, import a backup, review output, write a retro, score).
Binary subcommands (verification, no LLM loop)
Run these directly on any CLI directory + spec — they don't need the skill layer:
# Two-tier quality scorecard (Grade A = 85+): infrastructure + domain correctness
cli-printing-press scorecard --dir ./my-pp-cli --spec ./openapi.json
# Dogfood: catches dead flags, dead functions, auth mismatches, invalid paths
cli-printing-press dogfood --dir ./my-pp-cli --spec ./openapi.json
# Runtime verify against real API (read-only) or mock server
cli-printing-press verify --dir ./my-pp-cli --spec ./openapi.json --api-key "$TOKEN"
# Baseline snapshot for the improvement cycle
cli-printing-press emboss --dir ./my-pp-cli --spec ./openapi.json --audit-only
# Which library CLIs would benefit from new MCP spec surface
cli-printing-press mcp-audit
# Token diagnostics for printed CLIs (fingerprints only, never full tokens)
cli-printing-press auth doctor [--json]
Toolbelt (local augmentation CLIs)
Real augmentation tools installed in tools/ that climaker shells out to around the
press (never patched into the binary). Full versions, install commands, and per-stage
mapping: tools/README.md. The "why each matters" ranking: references/augmentation-registry.md.
| Tool | Shell out when… | Command |
|---|---|---|
har2spec.sh (wraps har-to-openapi 2.5.0) |
user has a .har/sniff capture but no docs — bridge into the press's --spec/--har path (Stage 2) |
tools/har2spec.sh capture.har openapi.json → then cli-printing-press scorecard --spec ./openapi.json or /printing-press --spec ./openapi.json |
| restish v2.2.0 | polishing a printed CLI's flags/auth (mine its response-shorthand + auth schemes), or smoke-testing a real API's shape (Stage 5) | restish --version · restish api configure <name> · restish <name> <op> -f 'body.items[].name' |
| ogen v1.22.0 | hardening the typed internal/client — generate a reference Go client to diff vs the press output (Stage 3) |
ogen --target ./ref-ogen --package ref openapi.json |
| oapi-codegen v2.7.1 | same Stage 3 cross-check, alt generator | oapi-codegen -generate types,client -package ref openapi.json > ref_client.go |
mark3labs/mcp-go and modelcontextprotocol/go-sdk are libraries, not CLIs — nothing to
install/run. They're vendored by a generated <api>-pp-mcp server and are PR targets for the
press's MCP template; consult the repos, don't shell out.
What you get from one run
Each /printing-press <api> run produces two binaries — <api>-pp-cli (Cobra CLI) and <api>-pp-mcp (MCP server) — sharing one internal/client, internal/store, and auth. Plus research docs, verification proofs, and a Quality Score.
- Agent-native flags on every command:
--json,--select,--dry-run,--stdin,--csv,--compact,--quiet,--yes,--no-input,--no-cache,--no-color. Auto-JSON when piped (no--jsonneeded). Typed exit codes:0success,2usage,3not-found,4auth,5API,7rate-limited. - Local-first data layer: domain SQLite tables (not JSON blobs), FTS5 search, incremental
syncwith cursors.syncpulls,searchfinds in ms,sqlqueries raw,export/tail. - Compound / insight commands (from domain archetype):
stale,orphans,load,channel-health,reconcile,health,similar,trends,bottleneck.
Where output goes
- Active runs:
~/printing-press/.runstate/<scope>/runs/<run-id>/working/<api>-pp-cli - Published CLIs:
~/printing-press/library/<api> - Archived manuscripts:
~/printing-press/manuscripts/<api>/<run-id>/(research/,proofs/,discovery/,pipeline/)
<scope> derives from the git checkout path so parallel worktrees don't collide. --output overrides the generated CLI location.
Three input modes
- API name → press resolves the OpenAPI spec (19 pre-verified APIs in catalog: Asana, Discord, GitHub, Stripe, Twilio, …).
- Website URL, no docs → browser-sniff gate launches a browser, captures traffic, reverse-engineers the spec.
--spec ./openapi.jsonor--har ./capture.har→ explicit spec or DevTools capture.
Guidelines
- Default to the press's own skills.
/climakerinstalls + routes; it does not duplicate the generation pipeline. After Preflight, hand off to/printing-press*. - Reload after a skills install before expecting the
/printing-pressslash commands to exist. - Legal: technical capability ≠ permission. Many services' ToS prohibit automated access. Confirm the user is authorized before sniffing or wrapping a service. (The press echoes this limitation; respect it.)
- Live verify is read-only (GET,
--limit 1, 10s timeout, stops on 401) — it never mutates. - Codex fallback is automatic: after 3 consecutive Codex failures, codegen falls back to local Opus, no intervention.
- Browser-sniff is manual capture — point a browser or import a HAR; the press doesn't crawl autonomously.
Reference
- Repo: https://github.com/mvanhorn/cli-printing-press
- Library / catalog: https://github.com/mvanhorn/printing-press-library · https://printingpress.dev
- Codex notes:
docs/CODEX.md· Pipeline contract:docs/PIPELINE.md - Augmentation registry:
references/augmentation-registry.md— ranked upstream repos (mcp-go, oapi-codegen, restish, har-to-openapi, …) mapped to the press pipeline stage each can strengthen. Consult when asked to make climaker/the press "stronger" or to source an integration.
04 同仓库更多 Skills
zi-search
暂无简介
visualise
Render inline interactive visuals — SVG diagrams, HTML widgets, charts, and explainers — directly in the conversation. Use this skill whenever the user asks to visualize, diagram, chart, illustrate, or explain something visually, or when an explanation would genuinely benefit from a spatial/interactive diagram rather than text. Also triggers for: flowcharts, architecture diagrams, data visualizations, interactive explainers, comparison layouts, UI mockups, and any request containing 'show me', 'draw', 'map out', 'visualize', or 'diagram'. Even when the user doesn't explicitly ask for a visual, use this skill proactively when the topic has spatial, sequential, or systemic relationships that a diagram would clarify better than prose.
video
When the user wants to create, generate, or produce video content using AI tools or programmatic frameworks. Also use when the user mentions 'video production,' 'AI video,' 'Remotion,' 'Hyperframes,' 'HeyGen,' 'Synthesia,' 'Veo,' 'Runway,' 'Kling,' 'Pika,' 'video generation,' 'AI avatar,' 'talking head video,' 'programmatic video,' 'video template,' 'explainer video,' 'product demo video,' 'video pipeline,' or 'make me a video.' Use this for video creation, generation, and production workflows. For video content strategy and what to post, see social-content. For paid video ad creative, see ad-creative.
manim-video
Production pipeline for mathematical and technical animations using Manim Community Edition. Creates 3Blue1Brown-style explainer videos, algorithm visualizations, equation derivations, architecture diagrams, and data stories. Use when users request: animated explanations, math animations, concept visualizations, algorithm walkthroughs, technical explainers, 3Blue1Brown style videos, or any programmatic animation with geometric/mathematical content.
video-use
Edit any video by conversation. Transcribe, cut, color grade, generate overlay animations, burn subtitles — for talking heads, montages, tutorials, travel, interviews. No presets, no menus. Ask questions, confirm the plan, execute, iterate, persist. Production-correctness rules are hard; everything else is artistic freedom.
ui-improver
暂无简介