# Show HN: Open-Source Origami CLI > Source: > Published: 2026-09-25 16:49:28+00:00 Every [v3](https://docs.origami.chat/v3/overview) operation is a command: campaigns, people, templates, leads, jobs, account. Generated from Origami's published OpenAPI spec, so it covers the whole API and stays current. [Install](#install) · [Quickstart](#quickstart) · [Commands](#v3-send-leads-jobs-account) · [Origami docs](https://docs.origami.chat/) · [Contributing](https://github.com/BrainGridAI/origami-cli/blob/main/CONTRIBUTING.md) Built by the team behind [BrainGrid](https://www.braingrid.ai/?ref=origami-cli), [Plansmith](https://www.plansmith.co/?ref=origami-cli) and [Radial](https://www.radial.build/?ref=origami-cli). - **The whole API.** All 120 v3 operations are commands, built from[Origami's OpenAPI spec](https://docs.origami.chat/openapi-v3.yaml) . When Origami ships a change,`pnpm sync:v3` regenerates the catalog; nothing is hand-maintained. - **Checks before it sends.** Enum values and required fields are validated before any request leaves your machine.`launch --dry-run` shows the exact blockers a real launch would hit. - **Safe to retry.** Every POST carries an`Idempotency-Key` that is reused across automatic retries, so a flaky network never launches a campaign or enrolls a list twice. - **Built for scripts.** JSON when piped, tables in a terminal, CSV on request.`--all` follows cursors to the end, and data goes to stdout while progress goes to stderr. - **Async done right.** Anything that returns a Job takes`--wait` , which polls on the Job's own`next_poll_at` .`origami jobs wait ` picks up any Job later. ``` npm install -g @braingrid/origami-cli # puts `origami` on your PATH npx @braingrid/origami-cli --help # or run it without installing ``` Requires Node.js 18+. From source: `pnpm install && pnpm build && npm link`. ``` origami auth login # paste an og_live_… key origami send campaigns list --status draft # what's waiting to go out origami send campaigns examples list # the copy each person will get origami send campaigns launch --dry-run # blockers + held-back counts origami send campaigns launch # go origami send campaigns stats ``` Create an API key under **Settings → Developers** in the [Origami app](https://origami.chat). Keys look like `og_live_…` and are **parent-wide** (they can act on your org or any of its projects). ``` origami auth login # prompts for a key (hidden input), verifies it origami auth login --key og_live_… # non-interactive echo "$KEY" | origami auth login # from a pipe / secret manager origami auth login --project # store a default project (child org) scope ``` Inspect or rotate: ``` origami auth status # resolved profile (key masked) + plan origami auth logout # forget the active profile's key ``` The key can also come from the environment (handy for CI): ``` export ORIGAMI_API_KEY=og_live_… origami account ``` Origami is agentic: you send a brief, an agent researches it in the background and builds a table, then you read the rows. ``` # 1. Run a brief (creates an agent, polls to completion, prints the table) origami run "Find 20 fintech CEOs in NYC" --model max # 2. If the agent asks a question, answer it (same agent + conversation) origami agents ask "Only Series A and later, please" # 3. Read the rows it built origami tables rows origami tables rows -o csv --out leads.csv # export to a file # 4. Bring your own data — upsert + enrich origami tables upsert --match domain \ --row '{"domain":"acme.com"}' --row '{"domain":"globex.com"}' origami enrichments get --wait # poll until settled ``` `origami run` and `origami agents create` are the same flow; `run` is the friendly shortcut. Add `--no-wait` to return the running run id instead of polling, then poll later with `origami agents run-get --wait`. Place these **after** the command (kubectl-style): | Option | Description | |---|---| | `--api-key ` | API key (overrides profile/env) | | `--project ` | scope the request to a project (child org) via `x-origami-project` | | `--profile ` | configuration profile | | `--base-url ` | override the API base URL | | `-o, --output ` | `table` (default in a terminal),`json` (default when piped), or`csv` | | `--fields ` | comma-separated columns to show (table/csv) | | `--no-color` | disable colored output (also honors `NO_COLOR` ) | | `--debug` | log HTTP requests to stderr | | `--timeout ` | per-request timeout | | `--max-retries ` /`--no-retry` | control automatic retries on 429/5xx | Data goes to **stdout**; status, progress, and hints go to **stderr** — so `| jq` and `-o csv` pipelines stay clean. ``` origami run "" [--model lite|max] [--workspace ] [--table ]... [--rows] [--no-wait] origami agents list [--search ] [--all] origami agents get origami agents ask "" origami agents runs origami agents run-get [--wait] [--include stats,transcript] origami agents tables origami agents cancel origami agents archive origami tables list [--workspace ] [--all] origami tables get [--include stats] origami tables columns origami tables rows [--filter ' [value]']... [--sort :] \ [--flat] [--no-defaults] [--limit <=200] [--all] [-o csv] [--out file.csv] origami tables row origami tables cell origami tables upsert --match (--row ''... | --file rows.csv) \ [--no-enrich] [--reenrich-updated] [--batch-id ] ``` Filter operators: `contains`, `not_contains`, `equals`, `not_equals`, `is_empty`, `is_not_empty`, `greater_than`, `greater_than_or_equal`, `less_than`, `less_than_or_equal`. ``` origami enrichments list [--table ] [--all] origami enrichments get [--wait] origami documents list origami documents upload [--mode table|append|document] [--table ] origami documents get [--out ] origami documents rename origami documents delete origami workspaces list origami workspaces create --name origami workspaces get origami workspaces delete [--confirm] ``` Each operation id is the command path: `send.campaigns.people.list` is `origami send campaigns people list`, `fetch_more` becomes `fetch-more`. A group whose only operation is `get` collapses into the group (`origami send campaigns stats `), and an argument-free `get` is its group's default (`origami account`, `origami account credits`). ``` # The send recipe, in order origami send campaigns create --name "Q4 delivery leaders" --channels linkedin origami send campaigns people upsert --list-id origami send campaigns templates put --data @template.json origami send campaigns settings patch --no-auto-lead-refill-enabled --require-message-approval origami send campaigns senders add --sender-id origami send campaigns examples list # read the rendered copy per person origami send campaigns launch --dry-run # blockers + held-back counts origami send campaigns launch origami send campaigns stats # Leads origami leads lists list origami leads lists rows list --all -o csv > rows.csv origami leads lists rows upsert --rows @rows.json --match-columns email origami leads searches create --brief "…" --count 25 --wait # Jobs: every 202 returns one; --wait on the command, or poll later origami jobs list --status running origami jobs wait origami jobs input --answers "Only Series A and later" # Account origami account senders list --channel linkedin origami account rate-limits origami account exclusion-lists people add --email someone@example.com ``` How flags map to the API: - Path parameters are positional arguments; query parameters and top-level body fields are `--kebab-case` flags.`origami --help` lists them with the spec's descriptions and ends with the HTTP call and its docs page. - Booleans take `--flag` or`--no-flag` (sends`false` ). Lists take commas or repeats (`--ids a,b --ids c` ). Object fields take JSON inline,`@file.json` , or`-` for stdin. - `--data ` supplies the whole body; flags override its keys. A body field that would shadow a global option gets a`body-` prefix (`--body-fields` ). - Enums and required fields are checked before any request is sent. - Every POST carries an `Idempotency-Key` (yours via`--idempotency-key` , else a UUID) that is reused across automatic retries, so a retried launch or enrollment never runs twice. - List commands print one page and hint the cursor; `--all` follows`next_cursor` to the end. - Operations that return a Job accept `--wait` (polls on the Job's`next_poll_at` , default ceiling 30 minutes,`--wait-timeout ` to change it). Regenerate the command catalog when Origami ships spec changes: ``` pnpm sync:v3 # fetches https://docs.origami.chat/openapi-v3.yaml → src/v3/catalog.ts pnpm sync:v3 ./spec.yaml # or from a local copy origami campaigns list (--workspace | --table ) origami campaigns create "" # agentic — drafts sequences, doesn't send origami campaigns edit "" # agentic origami campaigns people [--status ] [--all] origami campaigns stats origami campaigns launch [--dry-run] # also: pause / resume origami campaigns delete [--confirm] origami sequences list (--workspace | --table | --column ) [--status ] origami sequences get origami sequences stop [--dry-run] origami sequences delete [--force] origami scheduled list [--workspace ] [--enabled true|false] origami scheduled create --workspace --name --prompt "" --cron "0 9 * * 1" origami scheduled enable|disable|trigger origami scheduled runs origami projects list # parent-org only origami projects create [--monthly-credits ] origami projects update [--name ] [--monthly-credits ] origami projects delete [--confirm] origami credits # credit balance (v3 shortcut) ``` Origami webhooks are configured in the dashboard; these are local helpers for building a receiver. ``` origami webhooks verify --secret whsec_… --id --timestamp --signature --file body.json origami webhooks listen --port 9333 --secret whsec_… # verify + pretty-print inbound events origami config path origami config show # secrets redacted origami config profiles origami config use origami tables rows -o json | jq '.[].website' origami agents list --all -o csv > agents.csv origami run "Find 50 founders" --rows -o json > result.json ``` The client automatically retries `408/429/500/502/503/504` and network errors with exponential backoff (honoring `Retry-After`), and polls runs on the API's `Retry-After` cadence (15s). Rate limits: 300 req/min per IP, 100 req/min per org; concurrent agent runs are plan-tunable. | Code | Meaning | |---|---| | `0` | success | | `1` | API or runtime error | | `2` | usage / configuration error (bad flag, missing key) | | Variable | Description | |---|---| | `ORIGAMI_API_KEY` | API key (overrides the stored profile key) | | `ORIGAMI_PROJECT` | default project id ( `x-origami-project` ) | | `ORIGAMI_BASE_URL` | API base URL | | `ORIGAMI_PROFILE` | default profile name | | `ORIGAMI_CONFIG_DIR` | directory for the config file (default `~/.config/origami` ) | | `NO_COLOR` /`FORCE_COLOR` | disable / force colored output | ``` pnpm dev -- # run from source (tsx) pnpm typecheck # tsc --noEmit pnpm test # vitest pnpm build # tsup → dist/index.js pnpm sync:v3 # regenerate src/v3/catalog.ts from the published v3 spec ``` Issues and pull requests are welcome. [CONTRIBUTING.md](https://github.com/BrainGridAI/origami-cli/blob/main/CONTRIBUTING.md) covers setup, the test suite, and how the v3 catalog is regenerated. Please report security issues privately as described in [SECURITY.md](https://github.com/BrainGridAI/origami-cli/blob/main/SECURITY.md). [BrainGrid](https://www.braingrid.ai/?ref=origami-cli): the app builder that plans before it builds · [Plansmith](https://www.plansmith.co/?ref=origami-cli): the planning agent for the business side of software · [Radial](https://www.radial.build/?ref=origami-cli): the issue tracker for coding agents [MIT](https://github.com/BrainGridAI/origami-cli/blob/main/LICENSE). origami-cli is an independent project and is not affiliated with or endorsed by Origami. "Origami" is a trademark of its owner.