Open-source GTM workspace for founders across industries. Pay-per-result, signed receipts, founder-led discipline encoded. Terminal CLI + local web dashboard over one SQLite ledger.
oneshot-gtm.com · what a signed receipt is · docs
bunx oneshot-gtm-server # dashboard only — published, no clone
oneshot-launch-voiced.mp4
OneShot is a pay-per-use API toolbox — email, SMS, voice, deep research, person enrichment, browser automation, website build — settled per call in USDC on Base, with a cryptographically signed receipt for every action.
oneshot-gtm is the strategy wrapper. It encodes the founder-led-sales playbook — Mom Test, Sean Ellis 40%, Predictable Revenue, do-things-that-don't-scale, multichannel cadence, signed-receipt CAC — as named plays you run from the terminal or the dashboard. Finders watch public signals, an ICP gate filters what they find, you approve rows in a queue, plays draft and send, cadences follow up, and every dollar lands in a receipt you can attribute to an outcome.
MIT, so you can read every prompt, fork every play, and trust what's running.
| Them | oneshot-gtm | |
|---|---|---|
| Pricing | Seat-based SaaS, $$/seat/mo | Pay-per-result, no subscription |
| Source visibility | Closed; you trust the dashboard | MIT; read the prompts, fork the plays |
| CAC story | Blended, estimated, dashboard-shaped | Signed per-call receipts, exportable |
| PMF posture | Assumes PMF, optimizes sends | Pre-PMF aware, soft-gates on scale moves |
| LLM | Built-in, opaque | BYO key (OpenRouter / OpenAI / Anthropic) |
| State | Vendor cloud | Local SQLite + chmod-600 dotfile |
Most GTM tools assume you have product-market fit and optimize sends. Most pre-PMF founders don't, and end up scaling a broken motion because the tool said "send more" — the Startup Genome Report's top documented cause of startup death. So the discipline is built in:
- Plays use your founder voice and the buyer's context, with low volume (≤50/day). Choose the motion and review its cadence before sending.
- First touches lead with a supported observation; the short introduction fits around the thought rather than interrupting it. The Offer says the useful thing in the email, for free; the CTA asks for one line the reader can answer from their own experience, or asks for nothing. It never asks a stranger for a meeting. One true concession you write in config (
founderAdmission) is worked into roughly a third of first touches; leave it blank and the beat is skipped, never invented. - Every draft passes a lint built on the Wikipedia "Signs of AI writing" canon — banned phrases, em dashes, AI vocabulary, three-item lists, sycophantic openers.
- Scale-move commands (
handoff templatize,first-ae,readiness) print soft-gate checklists and default to "not yet, fix this first" until the signals earn the move.--forceoverrides. - Every paid action emits a signed receipt carrying a memo (why the call happened), structured
decisionContext, and agoalIdgrouping a cadence's spend. When a reply or deal outcome lands, that value is tagged back, so CAC and RoCS on the Measure page are attestable and outcome-attributed, not estimated.
Use the dashboard without writing code. Describe what you sell (a product or service), who buys it, and one useful observation you can share. Choose sources and plays for those buyers. Industry packs cover restaurants, trades, healthcare practices, auto services, professional services, trucking, government and developer tools; they are editable starting points, not eligibility requirements.
Start your first workspace walks through setup, choosing a motion and reviewing the first draft. Workspaces explains multiple businesses. Repository boundaries explains what stays private: SDK internals, credentials, prospect data and your operating workspace.
The current motions focus on reaching identifiable business buyers and learning from customers. Consumer advertising, retail checkout and a ready-made finder for every industry or country are not included. You can bring your own prospects when the built-in sources do not cover your market.
curl -fsSL https://bun.sh/install | bash # Bun is the required runtime
git clone https://github.com/oneshot-agent/oneshot-gtm
cd oneshot-gtm && bun install
bun run cli -- init # config + keys wizard
bun run cli -- doctor # sanity check
bun run --cwd apps/web build # one-time: build the SPA
bun run cli -- ui # http://127.0.0.1:3030init asks for the founder profile the prompts draw on: background that builds trust, products you've shipped, notable partners or customers, and one true concession. All optional — a blank field skips the beat that uses it rather than improvising. Edit any of them later from /setup or config founder. What reaches the model, and when maps each field to the surfaces it shapes. A voice card adds how you write, in your words, to every email draft.
Some finders need keys init never asks about — most people want GITHUB_TOKEN (a classic token with no scopes) before enabling the GitHub finders. /setup and config keys store them; finders lists them.
To call it from anywhere: cd apps/cli && bun link && bun link oneshot-gtm && cd -. Prefer the dashboard without cloning? bunx oneshot-gtm-server downloads and boots it. Bun is still required — the bundle uses bun:sqlite and Bun.serve, and fails loudly under plain node. The CLI itself is not published to npm.
Both read and write the same ~/.oneshot-gtm/ledger.sqlite.
bun run cli -- intel advise # interactive coach
bun run cli -- find watch --once # poll due triggers, enqueue candidates
bun run cli -- find drain podcast-guest --dry-run # preview approved /queue rows
bun run cli -- cadence advance # daily tick: poll inbox, fire follow-ups73 commands — bun run cli -- --help (or oneshot-gtm --help once linked) is the reference:
| Group | Commands |
|---|---|
init · doctor · ui |
setup wizard · health check · open the dashboard |
config |
llm · founder · keys · voice [--from <path>...] · linkedin-session · telemetry on|off · slack-webhook [url] · x-engine [engine] · spend-ceiling [amount|off] |
identities |
list · add · remove <id> — the sender pool |
gmail |
auth (OAuth a sending account) · placement (inbox-placement canary) |
smartlead |
connect — API key + pick Smartlead mailboxes into the pool (send-only) |
domains |
list · pause <domain> · resume <domain> — provisioned OneShot domains |
find |
watch · drain <play> · import --csv <file> --play <name> · enrich-linkedin · research-prospects · research-queue · research-products · synthesize-angles · score-prospects · backfill-fit-reason · calibrate — --fail-on-empty makes watch --once and drain exit 2 on a run that produced nothing |
motion |
post-funding concierge demo-no-show competitor-switch hiring-signal podcast-guest discovery-interview free-pilot — each takes --target <file>; breakup-revive reads the ledger |
cadence |
advance — poll inbound, fire due steps |
direct-mail |
list · upload · preview · refresh · approve · send · cancel · skip — individual physical-mail approvals |
discover |
icp interview-prep · icp synthesize · pmf classify · pmf survey · pmf survey-collect |
measure |
benchmark — compare this install's command activity with the opt-in telemetry cohort; supports --json |
intel |
advise · personalize · triage-replies · backfill-intent · weekly-review |
handoff |
readiness · templatize · first-ae |
demo |
seed · ui · reset — a fictional install for screenshots and video |
workspace |
list · create <name> · use <name> · current · path <name> · remove <name> — one isolated install per product; --workspace <name> on any command |
Spend, CAC, RoCS and outcome logging live in the dashboard's Measure and Cadences pages so there's one source of truth. Add --json to a read-only command (doctor, identities list, domains list, workspace list) for machine-readable output, or script the /api/measure/* routes.
bun run cli -- ui [--dev] [--port 4000] [--no-browser]Ten pages plus a run form:
- Home — spend, reply-rate trend, in-flight cadences, and a scheduler strip showing each trigger's state, last run and next due
- Queue — triggers table (enable, edit config, fire) plus the target queue with bulk approve and per-play Drain
- Prospects — search and browse every candidate a finder ever surfaced, any status, with who decided it and why, what happened after, and an override
- Add Prospect — paste a LinkedIn / X / GitHub URL; deep research builds a dossier, the LLM picks an angle against your ICP and drafts an intro, and the row lands in the queue
- Replies — every reply matched to its prospect, play and cadence status across all sender identities; answer in place, by hand or LLM-drafted and research-grounded
- Cadences — stop, log outcome, preview the next step, batch send email, and review/send direct mail per prospect
- Receipts — paginated, with the memo and value chip per call; click through to the signed payload
- Plays — cards with channel badges, a Run button, Copy CLI, and optional direct-mail steps
- Measure — CAC and RoCS by time range, plus per-cadence spend vs tagged value grouped by goal
- Setup — founder profile, ICP, product brief, LLM provider, wallet keys, sender identities, telemetry toggle
- Run a play (
/run/$playName) — editable target rows, dry-run toggle, drafts streamed back over SSE with lint flags and receipt links
A floating strategist dock on every page reads your ICP and product one-liner and proposes trigger configs as confirmation chips. Next to it, a privacy toggle masks names, emails, companies and phone numbers for screenshots. Demo mode seeds a fictional, fully-populated install for the same purpose.
Fifteen finders discover prospects, ICP-filter them, and enqueue into /queue for one-click approve or reject. Each runs as a trigger with its own interval and spend cap; the dashboard server runs the scheduler in-process, so enabling a trigger is enough.
| Finder | Signal |
|---|---|
show-hn |
same-day Show HN posts, via the HN Algolia API |
post-funding-auto |
funding announcements by ICP-derived industry × round |
accelerator-batch |
new cohorts — yc-oss directory, websearch fallback for Techstars / Antler / 500 / AI Grant |
job-change |
joined as <persona> announcements, filtered by persona and company |
hiring-signal |
open roles on Greenhouse / Lever / Workable / Ashby implying a need |
podcast-guest |
recent guests across Latent Space, Lenny's, 20VC, Acquired, Invest Like the Best |
github-topics |
repos by topic, then a manifest scan that detects the vendor stack deterministically — needs GITHUB_TOKEN |
github-stars |
recent stargazers of repos you watch; tag each repo competitor or adjacent to route the play — needs GITHUB_TOKEN |
luma-events |
upcoming events from Luma's city pages, gated per event by a topic + ICP check before any spend; pitches the hosts and featured guests |
gov-solicitation |
SAM.gov Sources Sought / Presolicitation notices by NAICS code via the SDK's govSolicitations; pitches the notice's published contracting officer, no lookup spend and no key |
civic-agenda |
city/county council agenda items via the Legistar/Granicus API, keyword-gated free before any paid call; pitches the meeting body's published contact |
breakup-revive |
your own ledger — prospects cold for 60–90 days. No LLM or OneShot spend |
x-reposters |
people who repost/quote X accounts you watch, in two lanes: builders who'd adopt (email cadence) and dev accounts with reach who'd boost a launch (one-touch email, or a hand-sent DM draft) — needs X API keys or TWITTERAPI_IO_KEY |
local-business |
main-street businesses via peopleSearch/companySearch (job title × industry × location × company size), or the SDK's localSearch places index with engine: local — routed to the free-pilot play |
local-registry |
newly-licensed main-street businesses over free public registries (Socrata business licenses, NPPES NPI, FMCSA Company Census), resolved to a domain with the SDK's localResolve — recent matches route to new-business, older to free-pilot |
All finders start disabled in a new workspace. Choose sources for your buyers or apply an industry pack; existing workspaces retain their saved settings. Two ICP gates run per candidate: a topic gate on the source, before any spend, and a person gate on the human's role, staged by cost and judging the role against your own ICP, without assuming technical skills or a founder title. Only a positive reject drops a candidate; rejections land in /queue as auditable rows you can override. Finders covers the prescreen, the gates, product research and review ordering.
Twenty-three of them. Thirteen have a Run page in the dashboard and drain from the queue:
show-hn · job-change · post-funding · accelerator-batch · hiring-signal · podcast-guest · competitor-switch · stack-consolidation · repo-interest · luma-events · sources-sought · civic-pilot · design-partner-loi
The last three are the institutional counterparts to the founder-to-founder register: sources-sought cites a specific SAM.gov notice and agency, civic-pilot cites a council agenda item, meeting date and a cooperative purchasing vehicle, and design-partner-loi walks an ask ladder — conversation, scoped pilot, non-binding LOI — for enterprise, government and hardware buyers, and refuses to draft at an owner-operator.
Six more drain from the queue without a Run form — profile-intro (what Add Prospect enqueues), breakup-revive, free-pilot, and the three x-reposters feeds: x-repost-intro, x-amplify, and x-amplify-dm, the one play that never auto-sends (it drafts DM text you send by hand, then Mark sent records the touch). concierge and demo-no-show are CLI-only because they open with a voice call and an SMS respectively; both are fed by trigger webhooks.
Most carry a cadence — a value follow-up, then a breakup, over roughly three to nine days, editable per play from /plays. Any reply, email or LinkedIn, stops every live cadence for that prospect. You can also stop one deliberately from /cadences with a reason: bad-timing stops become breakup-revive candidates after the cold window, not-a-fit and do-not-contact stay excluded.
LinkedIn — OneShot operates connected accounts, message sync, and authorized actions. GTM owns drafting and cadence decisions. Connect your account and manage messages from Replies. No GTM LinkedIn webhook setup is required for that integration. See LinkedIn.
An optional, individually approved physical-letter step. Plays → Direct mail picks eligible prospects per motion (a named person, a company, and a complete U.S. business address); Cadences → Review mail generates or uploads the letter, shows the print proof and price, and sends on approval. Bulk actions never send mailpieces. Direct mail has the selection rules and timing.
find watch --install-service generates a launchd agent (macOS) or systemd user unit (Linux) that keeps the watch daemon running; Windows schedules find watch --once. --fail-on-empty makes a clean-but-empty run exit 2, so a cron wrapper can tell idle from broken. config spend-ceiling caps what all finders and drains may spend per day; manual sends are never gated by it. Background monitoring has the service files, exit codes and ceiling semantics.
Outbound ships through a sender identity pool — any mix of OneShot wallet-owned sending domains, your own Gmail / Workspace accounts, and Smartlead-hosted mailboxes (send-only). Every email to a prospect comes from the identity that sent their first touch. New identities ramp 10/day, +10/week, to a 50 ceiling, per domain; when every identity is at cap, nothing sends over it — steps stay due until midnight. The inbox poll merges the OneShot inbox with every authorized Gmail account, so stop-on-reply works whichever identity sent, and replies from /inbox thread on both transports.
Bounces are harvested from Gmail mailboxes and classified; a hard bounce stops the cadence and suppresses the address. gmail placement sends one real canary between two of your mailboxes and reports where it landed, with the SPF/DKIM/DMARC verdicts — no DNS tooling. Sending and deliverability has the full rules.
One install is one product: one founder voice, one ICP, one product brief, one ledger, one sender pool. A fresh install is a single workspace named default. Selling a second product means a second workspace with its own profile, keys, identities and dashboard port:
bun run cli -- workspace create acme # its own home under ~/.oneshot-gtm-workspaces
bun run cli -- --workspace acme init # profile, keys, identities
bun run cli -- --workspace acme ui # side by side with the default dashboardWorkspaces share one thing: ~/.oneshot-gtm-shared/shared.sqlite, holding shared person identities, workspace memberships, paid lookup caches and contact touches, so resolved identities can reuse cached lookups and known email aliases share the 7-day cross-workspace first-touch hold. Unlinked aliases can still cause duplicate lookups. Workspaces covers resolution order, the shared DB, doctor's cross-workspace checks and the dashboard switcher.
┌──────────────────────────────────────────────┐
│ apps/cli apps/server apps/web │ ← surfaces
│ commander Bun.serve+SSE Vite+React │
└───────────────────────┬──────────────────────┘
│
┌───────────────────────┴──────────────────────┐
│ packages/* — the brains, shared by all 3 │
│ core · intel · plays · find · prompts · │
│ doctor · shared-types │
└───────────────────────┬──────────────────────┘
▼
┌──────────────────────────────────────────────┐
│ @oneshot-agent/sdk — OneShot primitives │
│ email · SMS · voice · research · enrichment │
│ browser · build · signed receipts │
└──────────────────────────────────────────────┘
State — one ~/.oneshot-gtm/ledger.sqlite is the source of truth for all three surfaces: receipts, prospects, sequence events, cadence state, deal outcomes, interviews, target queue, triggers, bounces, and sender assignments. ONESHOT_GTM_HOME relocates the whole directory.
Secrets — ~/.oneshot-gtm/.env, chmod 600, auto-loaded on first import.
Server — single-user, local-first, binds 127.0.0.1 only. Dashboard routes rely on that local boundary. Trigger webhooks accept signup and demo no-show events. OneShot messaging does not require a public GTM callback.
Data underneath — enrichment, verification and local lookups resolve through OneShot, which maintains a vendor landscape of 50 data sources across six categories (contact enrichment, company data, email verification, maps and places, browser automation, social). Which source answers a given lookup is chosen per call. The full catalogue.
apps/
cli/ the 73-command CLI (commander); src/demo/ seeds the demo install, src/main.ts picks the workspace
server/ Bun.serve + SSE; tsdown bundle published as `oneshot-gtm-server`
web/ Vite + React 19 + TanStack + Base UI — 9 pages, run form, strategist dock, privacy mode
packages/
core/ SDK wrapper, SQLite ledger, config + secrets, Gmail transport, JSONL events
intel/ LLM client, advise, personalize, triage, weekly-review
plays/ 23 outreach plays + handoff/icp/pmf modules + cadence engine
find/ 15 finders + shared pipeline (manifest scan, dedupe, ICP filter, drain, registry)
prompts/ Markdown prompts — humanizer canon, per-play, per-extract
doctor/ Wallet, ledger, key and deliverability health checks
shared-types/ Wire types shared across CLI / server / web
docs/ Guides: workspaces, finders, sending, direct mail, background monitoring, webhooks, demo mode, prompt inputs, voice
examples/ Sample target files for nine plays
Stack — Bun 1.3+ · Turborepo with a Bun catalog · Vitest 4 · oxlint + oxfmt · TypeScript 6 (verbatimModuleSyntax, noUncheckedIndexedAccess, noImplicitOverride) · Vite 8 + React 19 + TanStack Router/Query + Base UI + Tailwind 4 · tsdown for the server bundle · bun:sqlite · BYO LLM via OpenRouter, OpenAI or Anthropic. Plain async/await throughout — no monadic abstractions to learn before reading the code.
bun install
bun run typecheck # tsc --noEmit across cli + server + packages
bun run lint # oxlint
bun run fmt # oxfmt --write (fmt:check in CI)
bun --bun run test # vitest under Bun — current totals in STATUS.md
bun run cli -- doctor # smoke checkThe web app typechecks separately, because TanStack's file-based route tree needs a build step first:
bun run --cwd apps/web typecheck
bun run --cwd apps/web build # → apps/web/dist/
bun run --cwd apps/server build # → apps/server/dist/bin.mjs + dist/web/The suite is 4245 cases across 335 files (STATUS.md carries the current totals). Tests set ONESHOT_GTM_HOME to a temp dir, so they never touch your real ledger. CI runs bun --bun run test — the flag matters, since bun:sqlite doesn't exist under Node. Core and server install the pinned @oneshot-agent/sdk dependency from the package registry. SDK source and private archives do not belong in this repository. STATUS.md's "Last verified" line is generated on main — CI runs bun run status:stamp and commits the result on every push to main; never hand-edit that line in a branch, it will be overwritten on merge.
Every install writes a structured event log to ~/.oneshot-gtm/events.jsonl — one line per LLM call, ICP decision, finder lifecycle event and swallowed catch. Local-only, never transmitted; the ctx payload is primitives, counters, durations and hostnames only.
tail -F ~/.oneshot-gtm/events.jsonl | jq -c '{t:.ts, k:.kind, ctx:.ctx}' # condensed
tail -F ~/.oneshot-gtm/events.jsonl | jq -c 'select(.kind|startswith("llm."))' # LLM calls
tail -F ~/.oneshot-gtm/events.jsonl | jq -c 'select(.kind=="icp.decision")' # topic-gate rejects
tail -F ~/.oneshot-gtm/events.jsonl | jq -c 'select(.kind=="icp.person_decision")' # person-gate verdicts
tail -F ~/.oneshot-gtm/events.jsonl | jq -c 'select(.level=="error" or .level=="warn")'
tail -2000 ~/.oneshot-gtm/events.jsonl | jq -c 'select(.run_id=="PASTE-HERE")' # one run
DEBUG=oneshot:* oneshot-gtm find watch --once # mirror to stderrThe file rotates at about 10 MB into events.1.jsonl … events.3.jsonl (override with ONESHOT_GTM_MAX_EVENT_LOG_BYTES); tail -F survives the rotation, tail -f doesn't.
Anonymous, opt-out, one command to disable:
oneshot-gtm config telemetry off # or ONESHOT_GTM_TELEMETRY=0One summary event per invocation: command, flags, outcome, duration, version, OS. TELEMETRY.md is the authoritative field spec. Nothing about your prospects, prompts, replies, receipts or wallet leaves your machine.
STATUS.md lists what isn't yet proven against the live API. ROADMAP.md lists what isn't built — and, at the bottom, the things this deliberately will never do (run an SDR, manage your DNS, hold your customer data, lock you to an LLM, go multi-user).
MIT. See LICENSE.
Read every prompt. Fork every play. We expect you to.
Built by free.butter — the lead infrastructure behind this is the same pipeline that runs there.
