Agent-first command-line interface for the BikeBook Workshop Public API.
Status: v1 implementation in progress. See
RESEARCH.mdfor the design doc andLINEAR_STORIES.mdfor the implementation plan.
AI agents are the primary consumer of this CLI. The interface is built around
the agent-friendly conventions distilled in RESEARCH.md: structured JSON
output, deterministic exit codes, mandatory idempotency on writes, and a
machine-readable bikebook describe schema generated from the API's OpenAPI
spec.
# macOS / Linux
curl -fsSL https://raw.githubusercontent.com/helopony/bikebook-cli/main/install.sh | sh
# Homebrew
brew install helopony/tap/bikebook
# From source
go install github.com/helopony/bikebook-cli/cmd/bikebook@latestbikebook describe— full machine-readable schema of every command, flag, and exit code, generated frompublic-v1.json.bikebook --helpandbikebook <cmd> --help— per-command help.bikebook upgrade— self-update from the latest GitHub Release; setBIKEBOOK_NO_UPGRADE=1to disable.AGENTS.md— task-shaped command recipes for agents.llms.txt— curated Markdown link map for LLM tools..claude/skills/bikebook.md— Claude Code skill with install and common command patterns.
The full API spec lives at public-v1.json (OpenAPI 3.0.1,
48 operations across 13 resource tags). The CLI is a thin, faithful projection
of this spec — every subcommand maps to exactly one HTTP operation.