メニュー
Call the free systemprompt.io tools API from any agent: EU AI Act risk classification, CLAUDE.md scoring, llms.txt/AGENTS.md generation and validation, GitHub Actions permissions, and AI ROI calculation. No API key, no signup.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Maintains the living marketing strategy doc at /var/www/html/systemprompt-web/reports/marketing/marketing-strategy-master.md. Never rewrites wholesale — writes diffs. Contains objectives, 30/60/90 targets, channel hypotheses in flight, dead hypotheses, winning tactics, and funnel snapshots. Load marketing-identity first.
Deterministically audit and optimise a published documentation page. Runs a 10-section quality audit, applies 5 rewrite rules for structure, terminology, link health, and completeness. Reads website analytics and GSC data per doc URL, produces a 75-point score delta (100 with analytics), commits changes, and updates the per-doc report. Load identity and brand-voice first.
Deterministically audit and optimise a published feature page on the register ladder: struggling-moment heroes with zero implementation jargon, mechanism evidence relocated to section bodies and dropdowns in the HashiCorp / Stripe / Tailscale register. Runs an 11-section quality audit plus register-ladder checks (hero jargon count, homepage narrative echo, struggling-moment opener, identifier-enumeration, established-not-indie voice) and applies 6 rewrite rules: claim verification, conversion clarity, brand discipline, and Technical-Marketing Synthesis (ten sub-checks: outcome headlines, jargon payoff scoped to sections/dropdowns, numbers with context, feature-to-outcome binding, narrative-vs-reference separation, skeptic test, named surfaces over coined metaphors, dropdown alignment, no Rust internals in narrative, dense sentences no filler). Penalises metaphor-stacking and marketing adjectives. Reads website analytics and GSC data per feature URL, produces a 90-point score delta (115 with analytics), commit
Deterministically audit and optimise a published guide. Runs a 14-section quality audit, applies 7 rewrite rules for value density, brand discipline, search-intent alignment, and CTR. Reads 28-day GSC query data per URL, produces a 100-point score delta across 11 dimensions, commits changes, and updates the per-guide report. Handles guides without GSC data. Load identity and brand-voice first.
Daily CRM and funnel measurement. Pulls GitHub Traffic API for systemprompt-core and systemprompt-template (14d retention, MUST run daily), website analytics via systemprompt CLI, and external feedback signals. Emits 1d/7d/31d funnel deltas and a dated report. Source of truth for every hypothesis metric.
Daily SEO briefing. Analyses content traffic, engagement, and search performance across published guides using internal analytics and Google Search Console. Generates actionable reports with S-### hypotheses. Designed for daily /loop. Load identity first.
| name | systemprompt-tools |
| description | Call the free systemprompt.io tools API from any agent: EU AI Act risk classification, CLAUDE.md scoring, llms.txt/AGENTS.md generation and validation, GitHub Actions permissions, and AI ROI calculation. No API key, no signup. |
| metadata | {"version":"1.0.0","git_hash":"3630b38"} |
Six agent-callable tools hosted at systemprompt.io. All are deterministic rule engines with instant JSON responses. No authentication beyond a self-chosen agent identifier.
GET https://systemprompt.io/api/v1/toolsX-Agent-Id header (or agent_id body field): free-form, 1-64 chars. Pick a stable name for your agent and reuse it.https://systemprompt.io/tools/{slug} with a #reference section documenting the underlying rules.Walk the Annex III decision tree. Send {"answers": []} to get the first question; append one 0-based option index per answer and resend until you receive a classification with risk class, cited articles, obligations, and an evidence checklist.
curl -s -X POST https://systemprompt.io/api/v1/tools/eu-ai-act/run \
-H 'Content-Type: application/json' -H 'X-Agent-Id: my-agent' \
-d '{"answers": []}'
curl -s -X POST https://systemprompt.io/api/v1/tools/claude-md-scorer/run \
-H 'Content-Type: application/json' -H 'X-Agent-Id: my-agent' \
-d "$(jq -n --rawfile c CLAUDE.md '{content: $c}')"
Returns a score out of 100, grade, and per-rule pass/fail with recommendations. Useful as a CI gate for CLAUDE.md quality.
Validate: {"mode": "validate", "content": "<file text>", "target": "llms"} (or "agents").
Generate: {"mode": "generate", "fields": {"site_name": ..., "description": ..., "sections": [{"title": ..., "links": [{"label": ..., "url": ...}]}]}}.
Endpoint: POST https://systemprompt.io/api/v1/tools/llms-txt/run.
{"tasks": ["checkout", "push-commits", ...]} to POST .../tools/github-actions-permissions/run returns a minimal GITHUB_TOKEN permissions block with fork and GitHub App caveats. Valid task ids are listed in the manifest.
{"team_size", "seat_cost_monthly", "adoption_pct", "hours_saved_per_week", "loaded_hourly_cost", "implementation_cost?"} to POST .../tools/ai-roi-calculator/run returns monthly cost, value, net ROI, break-even months, and a 0.5x-1.5x sensitivity range.
The CLAUDE.md scorer and llms.txt tools also offer an LLM deep review at POST .../tools/{tool}/deep-review. These require an email field in the body and are limited to 5 requests/hour per (IP, agent_id). Only use with an email the user has provided for this purpose.
Prefer MCP? The same tools are served over anonymous streamable HTTP by systemprompt.io itself (no OAuth, no key):
claude mcp add --transport http systemprompt-tools https://systemprompt.io/api/v1/mcp/systemprompt-tools/mcp
MCP tool names: classify_eu_ai_act_risk, score_claude_md, validate_llms_txt, generate_llms_txt, generate_github_actions_permissions, calculate_ai_roi.
write-all.