Command line - Caplia Docs

phronel is a Node CLI (Node 18.17 or newer) with no runtime dependencies. It is the fastest way to give a coding agent access to Phronel: one login, then plain commands that print JSON whenever stdout is not a terminal.

npx phronel login                                     # paste a key from phronel.ai
npx phronel run monzo.com --wait                      # P15 market waves + decision report
npx phronel run "Acme Ltd" --deck ./deck.pdf --wait   # adds CRI readiness and thesis fit

Install it once with npm install -g phronel if you prefer phronel to npx phronel.

Commands

Command	Does
`login`, `whoami`	store the key (verified against the API) and show which key is in use
`run <company> [--deck f.pdf] [--deck-url u] [--webhook u] [--wait]`	start a run; `--wait` polls until complete or failed
`runs list`, `runs get <id> [--wait]`	runs, newest first; reading a run advances it
`scores <company_id>`	P15, CRI, thesis fit and the scoring jobs in flight
`report <company_id> [--html] [--out f] [--regenerate]`	the decision report as JSON or HTML
`deck add <company_id> f.pdf` or `--url u`	attach a deck; queues CRI and thesis fit
`theses list`, `theses create <name> --criterion "Team:4:…"`, `theses delete <id>`	theses for fit scoring
`credits`, `pricing`	allowance and the price ladder
`skill --install`	write an agent skill file into `.claude/skills/phronel/`

Conventions

Output is a table on a terminal and JSON when piped. --json forces JSON.
Errors go to stderr as {"error": {…}, "fix": "…"}; the fix line says what to do next.
Exit codes: 0 ok, 1 API error or a failed run, 2 usage, 3 authentication, 4 network.
PHRONEL_API_KEY and PHRONEL_API_URL override the stored config, which lives in ~/.config/phronel (or PHRONEL_CONFIG_DIR).
Test keys route to the sandbox host automatically; live keys to production.

For agents

phronel skill --install writes a skill file that teaches Claude Code (or any agent that reads .claude/skills/) when and how to call the CLI. The typical loop is run … --wait, then report … --html --out report.html for a human, or the JSON for the agent’s own reasoning.