> ## Documentation Index
> Fetch the complete documentation index at: https://docs.pav.bio/llms.txt
> Use this file to discover all available pages before exploring further.

# CLI

> The pav command: every endpoint from your terminal or an agent's shell.

`pav` is one static binary. Commands mirror the API: `pav <resource> list`
and `pav <resource> get`, with each query parameter as a flag.

## Install

```bash theme={"system"}
curl -LsSf https://pav.bio/install.sh | sh
```

Installs `pav` to `~/.local/bin`. On Windows:
`powershell -c "irm https://pav.bio/install.ps1 | iex"`.

## Authenticate

```bash theme={"system"}
pav auth login            # stores the key in your OS keychain
export PAV_API_KEY=...    # or use an environment variable
```

## List and get

```bash theme={"system"}
pav programs list --company MRNA --phase phase_3
pav recalls list --limit 5 --sort=-document_date
pav drug-applications get --application-key N:209637
```

Output is a table in a terminal and JSON when piped. `--format json|jsonl|csv|yaml`
picks one; `--query` takes a JMESPath expression:

```bash theme={"system"}
pav programs list --company MRNA --format json --query "data[].{drug: drug, phase: phase}"
```

## Every page

`--page-all` follows `next_cursor` (up to `--page-limit` pages, default 10):

```bash theme={"system"}
pav programs list --company PFE --page-all --page-limit 100 \
  --format jsonl | jq -c '.data[]? // .' > pfizer_programs.jsonl
```

## For agents

`--help` and `--schema` describe every command and parameter, so an agent in
a shell can discover the API as it goes. `pav generate-skills` writes a
`SKILL.md` per command for Claude Code and similar tools.