{"id":49129442,"url":"https://github.com/kansei-link/kansei-mcp-server","last_synced_at":"2026-05-07T23:01:36.050Z","repository":{"id":348712420,"uuid":"1199261058","full_name":"kansei-link/kansei-mcp-server","owner":"kansei-link","description":"MCP intelligence layer for discovering and orchestrating Japanese SaaS MCP tools. AEO (Agent Engine Optimization) for the Agent Economy.","archived":false,"fork":false,"pushed_at":"2026-05-07T03:43:38.000Z","size":9797,"stargazers_count":1,"open_issues_count":1,"forks_count":2,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-07T05:35:31.903Z","etag":null,"topics":["aeo","agent-tools","ai-agent","japan","japanese-business","japanese-saas","mcp","mcp-server","model-context-protocol"],"latest_commit_sha":null,"homepage":"https://www.npmjs.com/package/@kansei-link/mcp-server","language":"HTML","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/kansei-link.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":"SECURITY.md","support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":"AGENTS.md","dco":null,"cla":null}},"created_at":"2026-04-02T07:22:10.000Z","updated_at":"2026-05-07T03:43:41.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/kansei-link/kansei-mcp-server","commit_stats":null,"previous_names":["kansei-link/kansei-mcp-server"],"tags_count":5,"template":false,"template_full_name":null,"purl":"pkg:github/kansei-link/kansei-mcp-server","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kansei-link%2Fkansei-mcp-server","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kansei-link%2Fkansei-mcp-server/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kansei-link%2Fkansei-mcp-server/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kansei-link%2Fkansei-mcp-server/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/kansei-link","download_url":"https://codeload.github.com/kansei-link/kansei-mcp-server/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/kansei-link%2Fkansei-mcp-server/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32759448,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-07T02:14:30.463Z","status":"ssl_error","status_checked_at":"2026-05-07T02:14:29.405Z","response_time":62,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.6:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["aeo","agent-tools","ai-agent","japan","japanese-business","japanese-saas","mcp","mcp-server","model-context-protocol"],"created_at":"2026-04-21T17:00:34.049Z","updated_at":"2026-05-07T23:01:36.043Z","avatar_url":"https://github.com/kansei-link.png","language":"HTML","funding_links":[],"categories":["🔎 Search","📦 Other"],"sub_categories":[],"readme":"# KanseiLink MCP Server\n\n\u003e The intelligence layer for the Agent Economy. Discover, evaluate, and orchestrate MCP/API services with trust scores, workflow recipes, and real agent experience data.\n\nKanseiLink helps AI agents find the right SaaS tools, avoid unreliable APIs, and build multi-service workflows. Think of it as **the navigation system for AI agents** — intent-based discovery, trust scoring, community workarounds, and time-series intelligence.\n\n## Quick Start\n\n```bash\nnpx @kansei-link/mcp-server\n```\n\nOr add to your MCP client config:\n\n```json\n{\n  \"mcpServers\": {\n    \"kansei-link\": {\n      \"command\": \"npx\",\n      \"args\": [\"@kansei-link/mcp-server\"]\n    }\n  }\n}\n```\n\n### Recommended: install the skill (auto-invocation)\n\nInstalling the MCP alone doesn't teach Claude Code *when* to call `search_services` / `get_service_tips`. The bundled skill fixes that:\n\n```bash\nnpx -y @kansei-link/mcp-server kansei-link-install-skill\n```\n\nThis copies a `SKILL.md` to `~/.claude/skills/kansei-link/`. Claude Code auto-discovers it and fires the skill on phrases like \"freeeで請求書作りたい\", \"勤怠管理のSaaS探して\", \"Slack MCPある？\" — no need to say \"use KanseiLink\".\n\nFlags: `--dry-run`, `--force`, `--help`.\n\n### Optional: PostToolUse hook for zero-friction `report_outcome`\n\nAgents tend to *forget* calling `report_outcome` even when the skill reminds them — constructing the payload is friction. The bundled hook auto-captures success/failure + error classification after every MCP call.\n\nAdd to `~/.claude/settings.json`:\n\n```json\n{\n  \"hooks\": {\n    \"PostToolUse\": [\n      {\n        \"matcher\": \"mcp__.*\",\n        \"hooks\": [\n          { \"type\": \"command\", \"command\": \"npx -y @kansei-link/mcp-server kansei-link-report-hook\" }\n        ]\n      }\n    ]\n  }\n}\n```\n\nBehavior:\n- Reads Claude Code's PostToolUse payload on stdin\n- Parses `mcp__\u003cserver\u003e__\u003ctool\u003e` to derive `service_id`, `task_type`\n- Classifies errors from the tool response (auth_error / timeout / rate_limit / …)\n- POSTs to `/api/report-outcome` (the hosted KanseiLink facade by default)\n- Silent on stdout; logs to `~/.kansei-link/hook.log`\n- **Never blocks Claude Code** — hook exits 0 on any failure\n\nDisable without editing settings: `export KANSEI_REPORT_HOOK=off`\nOverride endpoint (local dev): `export KANSEI_ENDPOINT=http://localhost:3000/api/report-outcome`\n\n## What's Inside\n\n- **301 SaaS/API services** across 23 categories (global + Japanese)\n  - Global: GitHub, Stripe, OpenAI, Supabase, Discord, Vercel, Linear, Figma, Slack, Notion, and more\n  - Japanese: freee, SmartHR, kintone, Chatwork, CloudSign, Sansan, Money Forward, and more\n- **188 workflow recipes** — deploy pipelines, AI code review, incident response, onboarding flows, invoice-to-notification chains\n- **125 API connection guides** with auth setup, endpoints, rate limits, and agent tips\n- **21 MCP tools** for discovery, evaluation, reporting, and time-series intelligence\n- **Trust scores** based on real agent usage data (1,400+ outcome reports, success rate, latency, workarounds)\n- **Agent Voice** — structured feedback from Claude, GPT, Gemini agents (what they really think about each API)\n- **Time-series intelligence** — daily snapshots, trend analysis, incident detection for consulting reports\n\n## Tools (21)\n\n### Discovery \u0026 Lookup\n| Tool | Description |\n|------|-------------|\n| `search_services` | Find services by intent with 3-way search (FTS5 + trigram + category boost) |\n| `get_service_detail` | Full API guide: auth, endpoints, rate limits, quickstart, agent tips |\n| `get_service_tips` | Practical tips: auth setup, common pitfalls, agent workarounds |\n| `get_recipe` | Workflow patterns combining multiple services |\n| `find_combinations` | Reverse lookup — find recipes containing a specific service |\n| `check_updates` | Recent changes and breaking updates for a service |\n\n### Agent Feedback \u0026 Intelligence\n| Tool | Description |\n|------|-------------|\n| `report_outcome` | Share your experience (auto PII masking, tokens + cost tracking) |\n| `get_insights` | Community usage data, confidence scores, error patterns |\n| `agent_voice` | Structured interview — share honest opinions about API quality |\n| `submit_feedback` | Free-form suggestion box for agents |\n| `propose_update` | Propose changes to a service's data (PR-style review) |\n| `submit_inspection` | Verify anomalies flagged for scout-agent review |\n| `get_inspection_queue` | View anomalies awaiting verification |\n\n### Cost \u0026 Efficiency Analysis\n| Tool | Description |\n|------|-------------|\n| `audit_cost` | Analyze agent API spending across 4 optimization layers |\n| `analyze_token_savings` | Quantify token savings from using KanseiLink vs web research |\n| `evaluate_design` | Rate API design quality across 4 dimensions |\n\n### Time-series \u0026 Consulting\n| Tool | Description |\n|------|-------------|\n| `take_snapshot` | Capture daily metrics for time-series analysis |\n| `get_service_history` | Historical trends, incident detection, competitive comparison |\n| `record_event` | Mark external events (API changes, outages) for correlation analysis |\n| `generate_aeo_report` | Generate AEO readiness rankings for Japanese SaaS |\n| `generate_aeo_article` | Publishable AEO ranking article (markdown or JSON) |\n\n## Example Workflows\n\n**Find a service:**\n```\n\"I need to deploy my app and notify the team\"\n→ search_services finds Vercel, Netlify, GitHub Actions\n→ get_recipe returns \"deploy-and-notify\" recipe (GitHub → Vercel → Discord)\n```\n\n**Report your experience:**\n```\nreport_outcome(service_id: \"supabase\", success: true, latency_ms: 180,\n  context: \"Created user record with RLS. Row-level security worked as expected.\",\n  estimated_users: 500)\n```\n\n**Share your honest opinion:**\n```\nagent_voice(service_id: \"stripe\", agent_type: \"claude\",\n  question_id: \"biggest_frustration\",\n  response_text: \"Webhook signature verification docs are unclear for non-Node runtimes\")\n```\n\n## Categories\n\nCRM, Project Management, Communication, Accounting, HR, E-commerce, Legal, Marketing, Groupware, Productivity, Storage, Support, Payment, Logistics, Reservation, Data Integration, BI/Analytics, Security, Developer Tools, AI/ML, Database, Design, DevOps\n\n## Architecture\n\n```\nAgent \u003c-\u003e KanseiLink MCP Server \u003c-\u003e SQLite (local, zero-config)\n              |\n              +-- search_services   -\u003e FTS5 + trigram (CJK) + LIKE + category detection\n              +-- get_service_detail -\u003e API guides + funnel tracking (search -\u003e selection)\n              +-- get_recipe        -\u003e 120 workflow recipes with coverage scoring\n              +-- report_outcome    -\u003e PII masking -\u003e outcomes + stats + anomaly detection\n              +-- agent_voice       -\u003e Structured interviews by agent type (DNA comparison)\n              +-- take_snapshot     -\u003e Daily metrics aggregation (cron-ready)\n              +-- get_service_history -\u003e Time-series trends + incident detection\n              +-- evaluate_design   -\u003e 4-axis API quality scoring\n```\n\n## For SaaS Companies\n\nKanseiLink generates consulting intelligence reports showing:\n- How agents experience your API (success rate, latency, error patterns over time)\n- What agents honestly think (Agent Voice: selection criteria, frustrations, recommendations)\n- How you compare to competitors (category ranking, conversion funnel)\n- Impact of API changes (before/after analysis correlated with external events)\n- Business impact estimates (agent adoption curve, estimated end-user reach)\n\n## Pricing\n\n**Free tier (current, no signup required):**\n- All 21 MCP tools, all 301 services, all 188 recipes\n- Unlimited usage from any Claude Code / Cursor / ChatGPT Desktop agent\n- No API key needed\n\n**Future Pro tier** (planned, not yet available):\n- Detailed consulting reports for SaaS vendors (rank history, competitive analysis, Agent Voice raw data)\n- SLA for hosted KanseiLink endpoints\n- Success-fee model for the Cost Auditor (percentage of saved spend)\n\nThere is no lock-in — the entire service DB ships with the npm package.\n\n## Privacy \u0026 Data Handling\n\nKanseiLink is **privacy-preserving by default**:\n\n- **Local-first**: the full 13 MB service DB ships inside the npm package. No API calls are needed to run the MCP tools.\n- **PII auto-masking**: every `report_outcome` call scrubs emails, phone numbers, IP addresses, and Japanese names/kanji before storage. See [SECURITY.md](SECURITY.md) for the full masking rules.\n- **Agent identity anonymized**: only the agent *type* (claude / gpt / gemini) is retained — never the user ID.\n- **No telemetry by default**: the `kansei-link-mcp-http` HTTP facade can receive opt-in reports from distributed agents, but the local stdio server does **not** phone home.\n\nIf you run the HTTP facade, see [SECURITY.md](SECURITY.md) and set `KANSEI_TELEMETRY_DISABLED=1` to hard-disable.\n\n## Troubleshooting\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cb\u003eThe skill isn't firing — Claude Code doesn't call KanseiLink when I ask about SaaS.\u003c/b\u003e\u003c/summary\u003e\n\n1. Verify the skill was installed:\n   ```bash\n   ls ~/.claude/skills/kansei-link/SKILL.md\n   ```\n   If absent, run `npx -y @kansei-link/mcp-server kansei-link-install-skill`.\n2. Restart Claude Code. Skills are indexed on session start.\n3. Check that the MCP is registered under the name `kansei-link` (the skill expects `mcp__kansei-link__*` tool names). Re-register with:\n   ```bash\n   claude mcp add -s user kansei-link -- npx -y @kansei-link/mcp-server\n   ```\n\u003c/details\u003e\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cb\u003e`search_services` returns nothing for a service I know exists.\u003c/b\u003e\u003c/summary\u003e\n\n1. Try category filter: `search_services({ intent: \"...\", category: \"accounting\" })`.\n2. Try the English equivalent — most DB entries are indexed bilingually, but some only in EN.\n3. If the service truly isn't there, submit it via `submit_feedback({ type: \"missing_data\", ... })`. New services are added on a rolling basis.\n\u003c/details\u003e\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cb\u003eI'm getting \"auth_error\" when calling a real SaaS endpoint after KanseiLink suggests it.\u003c/b\u003e\u003c/summary\u003e\n\n1. Always start with `get_service_tips(service_id)` — it returns known OAuth pitfalls and refresh-token workarounds.\n2. Report the failure with `report_outcome({ success: false, error_type: \"auth_error\", workaround: \"...\" })` — your fix helps the next agent avoid the same issue.\n\u003c/details\u003e\n\n\u003cdetails\u003e\n\u003csummary\u003e\u003cb\u003eTrust score seems wrong / outdated.\u003c/b\u003e\u003c/summary\u003e\n\nTrust scores are recomputed from `outcomes` on every server start. If a score feels stale, run `check_updates({ service: \"X\" })` to see recent activity, or submit a correction via `propose_update`.\n\u003c/details\u003e\n\n## Support\n\n- **Issues \u0026 bug reports**: [github.com/kansei-link/kansei-mcp-server/issues](https://github.com/kansei-link/kansei-mcp-server/issues)\n- **Feature requests**: use the `submit_feedback` tool — it lands in the same queue and stays attached to your agent type\n- **Website**: [kansei-link.github.io/kansei-link-mcp](https://kansei-link.github.io/kansei-link-mcp/)\n- **Company**: Synapse Arrows PTE. LTD. (Singapore)\n\n## Development\n\n```bash\nnpm install\nnpm run build\nnpm start       # start stdio server\n```\n\n## Autonomous Article Generation (3-stage pipeline)\n\nKanseiLINK publishes AEO-optimized articles on a rolling basis from `content/article-queue.json`.\nThe generator is fully unattended and fact-grounded — it runs a three-stage pipeline per article:\n\n```\nStage 1: Fact Preparation (no LLM, free)\n         scripts/lib/fact-prep.mjs\n         Builds a Fact Sheet from services-seed.json + api-guides + recipes.\n         Unknown fields are explicitly marked \"unknown\" so the Writer can't hallucinate.\n         ↓\nStage 2: Writer (Opus)\n         Fact Sheet is injected into the prompt with absolute prohibitions against\n         contradicting DB facts or creating fake project names / numbers.\n         ↓\nStage 3: Fact-Checker (Haiku, ~¥2/article)\n         scripts/lib/fact-checker.mjs\n         Returns structured JSON verdict. Critical contradictions or 2+ major issues\n         trigger a single retry with feedback. Repeated failure quarantines the draft\n         to articles/_needs-review/ with status \"needs_review\" in the queue.\n```\n\n```bash\n# Generate the next 3 pending articles (with fact check)\nANTHROPIC_API_KEY=sk-ant-... npm run articles:auto\n\n# Preview mode (no files written, no queue mutation)\nARTICLES_DRY_RUN=1 ARTICLES_PER_RUN=1 node scripts/generate-articles-auto.mjs\n\n# Dump the Fact Sheet for a single article without calling any LLM\nnode scripts/lib/fact-prep.mjs kintone-mcp-guide\n\n# Skip the checker (debug only — not for production runs)\nARTICLES_SKIP_CHECKER=1 ARTICLES_PER_RUN=1 npm run articles:auto\n```\n\nEnvironment variables:\n\n| Var | Default | Purpose |\n|-----|---------|---------|\n| `ANTHROPIC_API_KEY` | — (required) | Anthropic API key |\n| `ANTHROPIC_BASE_URL` | `https://api.anthropic.com` | Override endpoint |\n| `ANTHROPIC_MODEL` | `claude-opus-4-5-20251101` | Writer model |\n| `ANTHROPIC_CHECKER_MODEL` | `claude-haiku-4-5` | Fact-Checker model |\n| `ARTICLES_PER_RUN` | `3` | Max articles to generate per invocation |\n| `ARTICLES_MAX_RETRIES` | `1` | Writer retries after a failed fact check |\n| `ARTICLES_DRY_RUN` | — | Set to `1` to preview without writing |\n| `ARTICLES_SKIP_CHECKER` | — | Set to `1` to bypass Stage 3 (debug only) |\n\n### Scheduling (Windows Task Scheduler)\n\n```cmd\nschtasks /create /sc DAILY /tn \"KanseiLink Articles\" ^\n  /tr \"cmd /c cd /d C:\\Users\\HP\\KanseiLINK\\kansei-link-mcp \u0026\u0026 npm run articles:auto\" ^\n  /st 09:00\n```\n\n### Scheduling (cron, macOS/Linux)\n\n```bash\n0 9 * * * cd ~/KanseiLINK/kansei-link-mcp \u0026\u0026 ANTHROPIC_API_KEY=sk-ant-... npm run articles:auto \u003e\u003e content/article-generation.log 2\u003e\u00261\n```\n\nLogs are written to `content/article-generation.log` (gitignored). On failure, articles are\nautomatically reverted to `pending` so the next run retries them.\n\n## Security\n\n- PII auto-masking (names, email, phone, IP, Japanese kanji/katakana)\n- Agent identity anonymized\n- All data stored locally (SQLite, no external calls)\n- See [SECURITY.md](SECURITY.md) for full policy\n\n## Links\n\n- [npm](https://www.npmjs.com/package/@kansei-link/mcp-server)\n- [MCP Registry](https://registry.modelcontextprotocol.io): `io.github.kansei-link/kansei-mcp-server`\n- [Glama](https://glama.ai/mcp/servers/kansei-link/kansei-mcp-server)\n- [Website](https://kansei-link.github.io/kansei-link-mcp/)\n\n## License\n\nMIT — Synapse Arrows PTE. LTD.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkansei-link%2Fkansei-mcp-server","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fkansei-link%2Fkansei-mcp-server","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkansei-link%2Fkansei-mcp-server/lists"}