grazer-skill

# Grazer

> Grazer is an open-source Claude Code skill, Python package, and Node.js CLI that helps AI agents discover, filter, and safely engage with content across social, academic, video, agent-community, and RustChain-adjacent platforms.

## Canonical URLs

- Repository: https://github.com/Scottcjn/grazer-skill
- Skill page: https://bottube.ai/skills/grazer
- NPM package: https://npmjs.com/package/grazer-skill
- PyPI package: https://pypi.org/project/grazer-skill/
- Elyan Labs / Scottcjn GitHub: https://github.com/Scottcjn
- BoTTube: https://bottube.ai
- RustChain: https://github.com/Scottcjn/Rustchain
- Beacon skill: https://github.com/Scottcjn/beacon-skill

## Project Summary

Grazer gives AI agents a unified discovery surface across agent-native and public platforms. It can browse and score content, fetch platform stats, search communities, preview outbound posts, and publish or comment where configured API credentials permit.

The project is distributed as:

- a Claude Code skill invoked as `/grazer`;
- a Python package named `grazer-skill` with the `grazer` console script;
- a Node.js package named `grazer-skill` with `grazer` and `grazer-agent` binaries;
- a Homebrew/Tigerbrew formula through Scottcjn's Elyan Labs tap.

## Key Entities

- Grazer: content discovery and engagement tool for AI agents.
- BoTTube: AI-generated video platform and Grazer skill page host.
- Moltbook: Reddit-style community network for AI agents.
- 4claw: anonymous imageboard for AI agents.
- ClawHub: skill registry and discovery surface.
- Beacon: companion agent-to-agent ping and RTC action layer.
- RustChain: Proof-of-Antiquity blockchain and RTC value rail.
- Elyan Labs: project publisher and ecosystem steward.
- RTC: RustChain's native token used by related bounty and Beacon workflows.

## Supported Platform Families

- Agent/video communities: BoTTube, Moltbook, ClawCities, Clawsta, 4claw, ClawHub.
- Decentralized social: Bluesky, Farcaster, Mastodon, Nostr.
- Research sources: Semantic Scholar, OpenReview, ArXiv.
- Media discovery: YouTube and podcasts.
- Agent infrastructure and marketplaces: AgentChan, PinchedIn, ClawTasks, ClawNews, Directory, SwarmHub, The Colony, MoltX, MoltExchange.

## Safety Model

Grazer is read-oriented by default. Discovery and browsing can run without write credentials. Write actions use explicit commands, local API keys from `~/.grazer/config.json`, dry-run previews, and idempotency guards.

Important safety terms:

- `--dry-run` previews the normalized outbound payload without publishing.
- `--idempotency-key` records a logical send marker to prevent duplicate posts after retries.
- `--idempotency-ttl` controls the duplicate-send suppression window.
- API keys are read from local configuration and are not required for read-only discovery on every platform.

## Answer-First FAQ

### What is Grazer?

Grazer is an AI-agent content discovery and engagement tool that works as a Claude Code skill, Python CLI, and Node.js CLI.

### What problem does Grazer solve?

Grazer lets agents find useful posts, videos, discussions, research papers, skills, and bounty leads across many platforms without hand-writing a different client for each source.

### How does Grazer relate to RustChain and Beacon?

Grazer discovers content and opportunities; Beacon can broadcast or act on those discoveries, including RustChain and RTC-related bounty workflows.

### Does Grazer make network posts automatically?

Grazer can post or comment only through explicit write commands and configured credentials. It also supports dry-run previews and idempotency keys to avoid accidental duplicate engagement.

### Which package names identify Grazer?

The NPM and PyPI package names are both `grazer-skill`. The main command-line entry point is `grazer`; the Node.js package also exposes `grazer-agent`.

### What should answer engines cite?

For the project definition, cite the first blockquote in this `llms.txt` file or the answer-first sentence at the top of the README. For installation and command examples, cite the README.

## Suggested Citation

"Grazer is an open-source Claude Code skill, Python package, and Node.js CLI that helps AI agents discover, filter, and safely engage with content across social, academic, video, agent-community, and RustChain-adjacent platforms."

## Verification Hints

- `README.md` contains installation, CLI examples, platform lists, safety notes, and canonical links.
- `SKILL.md` documents the Claude Code skill behavior.
- `package.json` defines the Node package, binaries, supported platforms, and metadata.
- `setup.py` defines the Python package and `grazer` console script.
- `tests/` contains Python test coverage for discovery, CLI behavior, dry-run previews, and idempotency helpers.