{"_id":"grep-research-skills","_rev":"2-76a54864ec67faeaa065be60bf059b1e","name":"grep-research-skills","dist-tags":{"latest":"0.2.0"},"versions":{"0.1.0":{"name":"grep-research-skills","version":"0.1.0","keywords":["grep","research","deep-research","claude-code","claude-code-skill","cowork","openclaw","ai-agent","claude"],"author":{"url":"https://grep.ai","name":"Parcha Labs"},"license":"MIT","_id":"grep-research-skills@0.1.0","maintainers":[{"name":"aj-parcha","email":"Aj@grep.ai"}],"homepage":"https://grep.ai","bugs":{"url":"https://github.com/parcha-ai/grep-research-skills/issues"},"bin":{"grep-research-skills":"bin/install.js"},"dist":{"shasum":"a02cfa1aee1e19fe8e2c3ae7fc6882717cf67a98","tarball":"https://registry.npmjs.org/grep-research-skills/-/grep-research-skills-0.1.0.tgz","fileCount":17,"integrity":"sha512-ty3tBZayDUefScD6UlpjChy1t6KG9YREg0Kx4giSg1kZvcQFaKhQsblxkpb9aW+0KAvKj7Mc+xdKWt3Tvw7IjA==","signatures":[{"sig":"MEQCIDwq8fNoLWMAVajEPEXOsGRTTZVECttOZhlKMK6TK6MRAiBGk8JXL+cTEAbShUKrPisvnD+N0s02K1pQuBeXuCVMAA==","keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U"}],"unpackedSize":112039},"engines":{"node":">=18"},"gitHead":"fc357ee3fe6bb964bb93ccd7854487ba85e11900","_npmUser":{"name":"aj-parcha","email":"Aj@grep.ai"},"repository":{"url":"git+https://github.com/parcha-ai/grep-research-skills.git","type":"git"},"_npmVersion":"11.4.2","description":"Give your AI agent deep research superpowers. Connects Claude Code, Cowork, and OpenClaw to GREP — the #1 deep research engine.","directories":{},"_nodeVersion":"24.3.0","_hasShrinkwrap":false,"_npmOperationalInternal":{"tmp":"tmp/grep-research-skills_0.1.0_1777489322002_0.05820426858274019","host":"s3://npm-registry-packages-npm-production"}},"0.2.0":{"name":"grep-research-skills","version":"0.2.0","description":"Give your AI agent deep research superpowers. Connects Claude Code, Cowork, and OpenClaw to GREP — the #1 deep research engine.","author":{"name":"Parcha Labs","url":"https://grep.ai"},"license":"MIT","repository":{"type":"git","url":"git+https://github.com/parcha-ai/grep-research-skills.git"},"homepage":"https://grep.ai","keywords":["grep","research","deep-research","claude-code","claude-code-skill","cowork","openclaw","ai-agent","claude"],"bin":{"grep-research-skills":"bin/install.js"},"engines":{"node":">=18"},"scripts":{"build:cowork":"node bin/build-cowork-zip.js"},"devDependencies":{"archiver":"^7.0.1"},"_id":"grep-research-skills@0.2.0","gitHead":"923f4b599badc9122a42ea353d671be373c07b4c","bugs":{"url":"https://github.com/parcha-ai/grep-research-skills/issues"},"_nodeVersion":"24.3.0","_npmVersion":"11.4.2","dist":{"integrity":"sha512-yo7V4IkaexFKzTmXKXA7fJVMWV5R+5hr54oi0HiBpqyxR6JeMnbwxX6R8hXcqzDKzkKYVSok+POiijAQwoib9w==","shasum":"c4d05807588268697fe82622dd52cbc2b40fd8e6","tarball":"https://registry.npmjs.org/grep-research-skills/-/grep-research-skills-0.2.0.tgz","fileCount":34,"unpackedSize":257893,"signatures":[{"keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U","sig":"MEUCIQDKFitWr9wALQ0ARpLAycf9a7b3Rchns0AaGz+vICZHUQIgfAM9XShRkep8KQaWsDKCoC8Kit0DKDIL4kQjYmNel38="}]},"_npmUser":{"name":"aj-parcha","email":"Aj@grep.ai"},"directories":{},"maintainers":[{"name":"aj-parcha","email":"Aj@grep.ai"}],"_npmOperationalInternal":{"host":"s3://npm-registry-packages-npm-production","tmp":"tmp/grep-research-skills_0.2.0_1785859886276_0.22669316600832934"},"_hasShrinkwrap":false}},"time":{"created":"2026-04-29T19:02:02.001Z","modified":"2026-08-04T16:11:26.555Z","0.1.0":"2026-04-29T19:02:02.170Z","0.2.0":"2026-08-04T16:11:26.401Z"},"bugs":{"url":"https://github.com/parcha-ai/grep-research-skills/issues"},"author":{"name":"Parcha Labs","url":"https://grep.ai"},"license":"MIT","homepage":"https://grep.ai","keywords":["grep","research","deep-research","claude-code","claude-code-skill","cowork","openclaw","ai-agent","claude"],"repository":{"type":"git","url":"git+https://github.com/parcha-ai/grep-research-skills.git"},"description":"Give your AI agent deep research superpowers. Connects Claude Code, Cowork, and OpenClaw to GREP — the #1 deep research engine.","maintainers":[{"name":"aj-parcha","email":"Aj@grep.ai"}],"readme":"# GREP Research Skills\n\nGive your AI agent deep research superpowers. GREP Research Skills connects Claude Code, Cowork, and OpenClaw to [GREP](https://grep.ai) — the #1 deep research engine.\n\n> The wallet/PAYG gateway path (Stripe Link, Base USDC, `Receipt pi_xxx` auth, x402 funding) lives on the [`mpp-gateway-future`](https://github.com/Parcha-ai/grep-research-skills/tree/mpp-gateway-future) branch. It returns once the backend's MPP gateway is GA. Until then this repo targets the v2 API only — Descope JWT or `grp_*` API key.\n\n## Install\n\n```bash\nnpx grep-research-skills\n```\n\nThat's it. Works with Claude Code, Cowork, and OpenClaw — the installer auto-detects your environment.\n\n**Requirements:** Node.js 18+\n\n### Alternative: Cowork / Claude.ai\n\nFor Cowork and Claude.ai, GREP is packaged as a **single `/research` skill** (instead of 8 individual skills). One router skill handles everything — it reads the right reference file based on your intent.\n\n1. Download the latest zip from [Releases](https://github.com/parcha-ai/grep-research-skills/releases/latest)\n2. In Cowork, go to **Settings → Plugins** and click **Add Plugin**\n3. Upload `grep-research-skills-v0.2.0.zip`\n4. The `/research` skill appears in your org — all team members get access\n\n**Why one skill?** Claude Code discovers skills as individual symlinks, so 16 skills = better auto-triggering. Cowork/Claude.ai loads skills from a zip, where a single well-described router skill with reference files is more practical and avoids cluttering the skill list.\n\n**Building the zip from source:**\n\n```bash\nnpm install            # one-time, pulls archiver (the only devDependency)\nnpm run build:cowork   # → dist/grep-research-skills-v0.2.0.zip\n```\n\nPure-JS — no system `zip` binary required. Works on macOS, Linux, and Windows.\n\n**Required: Allow network access.** In your Cowork org settings under **Code execution → Allow network egress**, add these domains to the allowlist:\n\n```\napi.grep.ai\npreview-api.grep.ai\napi.descope.com\n```\n\n(Add `preview-api.grep.ai` only if you're testing against the preview deployment via `GREP_API_BASE`.)\n\nWithout this, the sandbox can't reach the GREP API or authenticate via Descope.\n\n### Alternative: Claude Code Plugin Marketplace\n\n```\n/install-plugin parcha-ai/grep-research-skills\n```\n\n### Alternative: Git Clone\n\n```bash\ngit clone https://github.com/parcha-ai/grep-research-skills.git ~/.grep-research-skills && ~/.grep-research-skills/setup\n```\n\n## What You Get\n\n### Claude Code (16 individual skills)\n\n| Skill | Triggers on | What it does |\n|---|---|---|\n| `/grep-login` | \"log in to grep\", \"grep auth\" | Authenticate via email OTP or API key |\n| `/grep-status` | \"grep status\", \"what plan am I on\" | Account status + recent jobs |\n| `/grep-upgrade` | \"upgrade grep\", \"buy more credits\" | Choose / change subscription plan (Free / Pro / Ultra / PAYG) |\n| `/grep-skill-creator` | \"make a grep skill\" | Create a new SKILL.md for any agent skill, powered by deep research |\n| `/grep-plan` | \"plan a grep research\", \"research-informed plan\" | Research best practices + your codebase context before you `/plan` |\n| `/quick-research` | \"fast research / lookup\" | ~25s sourced one-liner — version checks, API endpoints, quick lookups |\n| `/research` | \"research X\" (default tier) | ~5 min comprehensive report with citations — the default for most tasks |\n| `/ultra-research` | \"deep / exhaustive research on X\" | Up to 1hr investigation — security audits, legal, ecosystem surveys |\n| **`/grep-mcp`** | \"install grep MCP\", \"grep as MCP server\" | Wire Grep into `.mcp.json` as 4 native MCP tools |\n| **`/grep-domain-expert`** | \"use the legal/medical/patent/... expert\" | Route to one of 27 public domain experts (legal, medical, financial, real estate, supply chain, maritime, etc.) |\n| **`/grep-build-app`** | \"build me an interactive app for X\" | Interactive HTML web apps via the app-builder expert (effort=build, ~$2, 10-15min) |\n| **`/grep-build-slidedeck`** | \"make me a slidedeck about X\" | Research-backed HTML slidedeck with arrow-key nav + PDF export |\n| **`/grep-build-spreadsheet`** | \"build a spreadsheet of X\" | Sortable HTML spreadsheet with CSV export |\n| **`/grep-research-workflow`** | \"investigate X and then make Y\" | Multi-step chain: orient → deep dive → optional build artifact |\n| **`/grep-with-context`** | \"research using these PDFs\" | Upload files as research inputs (PDFs, CSVs, images) |\n| **`/grep-continue`** | \"follow up on that research\" | Continue an existing job with a new question, inheriting prior research context |\n\n**Bold rows are new in 0.2.0.**\n\n### Cowork / Claude.ai (single consolidated skill)\n\n| Skill | Description |\n|-------|-------------|\n| `/research` | All features in one skill — routes to the right workflow (deep, quick, ultra, plan, skill-creator, login, upgrade, status) based on what you ask for |\n\n## Getting Started\n\n1. **Install** using `npx grep-research-skills`\n2. **Authenticate** by running `/grep-login` in your AI agent\n3. **Research** anything with `/research \"your topic\"`\n\n**Pick the right tier.** `/quick-research` is for one-liner answers, `/research` is the default for most tasks, and `/ultra-research` is reserved for heavy investigations that genuinely need exhaustive coverage (and can take up to an hour).\n\nFor domain-specific work (legal, medical, patent, etc.), reach for `/grep-domain-expert`. For deliverables (decks, apps, spreadsheets), use the matching `/grep-build-*` skill. For multi-step research with a deliverable at the end, use `/grep-research-workflow`.\n\n## Authentication (v2)\n\nAll requests use `Authorization: Bearer <token>` where `<token>` is either a Descope-managed session JWT or a long-lived `grp_*` API key. Both are stored in `~/.grep/session.json` (mode 0600) and read transparently by `scripts/grep-api.js`.\n\n```bash\n# Email OTP (interactive — best for human users)\n/grep-login\n\n# API key (headless / CI — best for agents and scripts)\nnode ~/.grep-research-skills/scripts/auth.js set-api-key grp_xxx\n```\n\nSessions auto-refresh. Get a `grp_*` key at https://grep.ai/api-keys. Bills against your subscription tier (Free / Pro / Ultra / PAYG) — run `/grep-status` to see your current plan and remaining quota.\n\n### Pointing at preview / staging\n\nSet `GREP_API_BASE` to override the default `https://api.grep.ai`:\n\n```bash\nexport GREP_API_BASE=https://preview-api.grep.ai   # early-access / staging\n# OR\nexport GREP_API_BASE=https://api.grep.ai           # production (default)\n```\n\nThe script auto-derives the matching UI host for printed report links (`api.grep.ai → grep.ai`, `preview-api.grep.ai → preview.grep.ai`). Override with `GREP_UI_BASE` if your deployment uses a different pattern.\n\n## Direct CLI Use\n\n```bash\nnode scripts/grep-api.js experts                            # 27-expert catalog (free, no auth)\nnode scripts/grep-api.js run \"What is Anthropic?\" --effort=low\nnode scripts/grep-api.js run \"compare LLM costs\" --output-type=spreadsheet --max-wait=1800\nnode scripts/grep-api.js files <slug>                       # workspace files for a job\nnode scripts/grep-api.js timeline <slug>                    # message timeline\nnode scripts/grep-api.js continue <slug> \"follow-up\"\nnode scripts/grep-api.js upload report.pdf                  # returns attachment_id\n```\n\n## Discovery\n\nThe v2 API publishes its contract via OpenAPI. Set `GREP_API_BASE` once and use it everywhere:\n\n```bash\nexport GREP_API_BASE=https://api.grep.ai             # production\n# OR\nexport GREP_API_BASE=https://preview-api.grep.ai     # preview / staging\n\ncurl \"$GREP_API_BASE/openapi.json\"                       # v2 OpenAPI\ncurl \"$GREP_API_BASE/api/v2/experts\"                     # 27-expert list (free, no auth)\ncurl \"$GREP_API_BASE/.well-known/agent-onboarding.md\"    # markdown agent guide\n```\n\n**Agents bootstrapping cold** should fetch `$GREP_API_BASE/.well-known/agent-onboarding.md` first — it describes the v2 surface end-to-end with copy-paste examples.\n\nThis repo also publishes a [skill manifest](.well-known/skill-manifest.json) for agent discovery.\n\n## How It Works\n\nGREP Research Skills uses headless email authentication (powered by Descope) — no browser needed. Works in terminals, SSH sessions, and headless environments.\n\n**`/quick-research` and `/research`** are blocking: the skill submits the job, polls with backoff, and returns the finished report in a single call. Claude Code's bash tool caps at 10 minutes, so `/research` is bounded to a 9-minute server-side wait. If a deep job overshoots, the skill exits with a `slug` for later retrieval.\n\n**`/ultra-research` is different.** Ultra-deep jobs can run up to 1 hour, which exceeds the bash 10-minute cap. The skill submits the job, returns the `slug` immediately, and polls on 5-minute intervals across multiple agent turns. You can keep working while it runs; the agent checks back periodically and presents the report when ready.\n\n**`/grep-build-*` and `/grep-with-context`** use Monitor with longer timeouts (up to 30 min for build jobs).\n\n## Authentication\n\n```bash\n# Authenticate (sends code to your email)\nnode ~/.grep-research-skills/scripts/auth.js login you@email.com\n\n# Check status\nnode ~/.grep-research-skills/scripts/auth.js status\n\n# Get token (for scripting)\nnode ~/.grep-research-skills/scripts/auth.js token\n```\n\nSessions are stored in `~/.grep/session.json` and auto-refresh.\n\n## For OpenClaw Users\n\nThe installer auto-detects OpenClaw and creates symlinks in `~/.openclaw/skills/`.\n\nYou can also publish to ClawHub or install skills manually:\n\n```bash\n# Drop into OpenClaw's skill directory\ncp -r ~/.grep-research-skills/skills/* ~/.openclaw/skills/\n```\n\n## Project Structure\n\n```\ngrep-research-skills/\n├── .claude-plugin/\n│   ├── plugin.json                # Claude Code plugin manifest\n│   └── marketplace.json           # Claude Code marketplace listing\n├── .well-known/\n│   └── skill-manifest.json        # Agent discovery manifest\n├── .github/\n│   └── workflows/\n│       └── sync-experts.yml       # Nightly drift check vs live /api/v2/experts\n├── skills/                        # Claude Code: 16 individual skills\n│   ├── research/SKILL.md          # Deep research (effort=medium, ~5 min)\n│   ├── quick-research/SKILL.md    # Fast fact check (effort=low, ~25s)\n│   ├── ultra-research/SKILL.md    # Exhaustive research (effort=high, up to 1 hr)\n│   ├── grep-plan/SKILL.md         # Research-informed planning\n│   ├── grep-skill-creator/SKILL.md  # Research-powered skill generator\n│   ├── grep-login/SKILL.md        # Authentication\n│   ├── grep-upgrade/SKILL.md      # Plan selection & Stripe checkout\n│   ├── grep-status/SKILL.md       # Status & job checking\n│   ├── grep-mcp/SKILL.md          # Wire Grep into .mcp.json (NEW in 0.2.0)\n│   ├── grep-domain-expert/SKILL.md  # Route to a public expert (NEW)\n│   ├── grep-build-app/SKILL.md    # Interactive HTML apps (NEW)\n│   ├── grep-build-slidedeck/SKILL.md  # HTML slidedecks (NEW)\n│   ├── grep-build-spreadsheet/SKILL.md  # Sortable spreadsheets (NEW)\n│   ├── grep-research-workflow/SKILL.md  # Multi-step chains (NEW)\n│   ├── grep-with-context/SKILL.md   # Research with attached files (NEW)\n│   └── grep-continue/SKILL.md     # Continue existing jobs (NEW)\n├── dist/cowork/                   # Cowork/Claude.ai: single consolidated skill\n│   └── research/\n│       ├── SKILL.md               # Router — routes intent to reference files\n│       ├── references/            # Workflow details for each intent\n│       │   ├── deep.md\n│       │   ├── quick.md\n│       │   ├── ultra.md\n│       │   ├── plan.md\n│       │   ├── skill-creator.md\n│       │   ├── login.md\n│       │   ├── upgrade.md\n│       │   └── status.md\n│       └── scripts/               # Bundled at build time by build-cowork-zip.js\n├── resources/\n│   ├── experts.md                 # 27-expert catalog (drift-tracked)\n│   ├── intent_map.md              # Phrase → API fields lookup\n│   ├── slidedeck_schema.json      # JSON Schema for structured deck output\n│   ├── spreadsheet_schema.json    # JSON Schema for structured table output\n│   └── chaining_examples.md       # Multi-step workflow recipes\n├── scripts/\n│   ├── auth.js                    # Descope OTP headless auth\n│   ├── grep-api.js                # GREP API client (v2)\n│   ├── billing.js                 # Billing & Stripe checkout client\n│   └── update-check.js            # Plugin auto-update\n├── bin/\n│   ├── install.js                 # npx installer\n│   └── build-cowork-zip.js        # Builds the Cowork/Claude.ai zip release\n├── setup                          # Shell installer (git clone fallback)\n├── package.json\n└── README.md\n```\n\n## License\n\nMIT — Parcha Labs, Inc.\n","readmeFilename":"README.md"}