Cargo CDK cookbooks

Each cookbook is a folder of define* resources that combine 3+ CDK resource types into one applyable outcome. They compose by sitting in the same project root: every cookbook builds on base-gtm (the shared accounts/contacts schema and connector set), so two cookbooks never collide and their resources stack.

Using a cookbook

Scaffold one directly (requires @cargo-ai/cdk with init --from support):

cargo-ai cdk init my-tam --from getcargohq/cargo-cookbooks/tam-building

--from reads cargo.scaffold.json and pulls the cookbook plus its required siblings (base-gtm, transitively) and the shared root files, keeping the folder layout so cross-folder imports resolve.

Or manually: keep base-gtm/ plus the cookbook folders you want; delete the rest. Then:

npm install            # postinstall syncs types if you are logged in (optional)
cp .env.example .env   # fill in the secrets the cookbooks you deploy need
cargo-ai cdk plan      # dry run: review the resource graph
cargo-ai cdk deploy

Run from this directory — defineContext paths are root-relative.

Conventions

• Placeholders are marked PLACEHOLDER in code comments — edit them before deploying (API keys via env, channel IDs, member uuids, persona filters).
• Shared resources (Slack, LinkedIn, waterfall, Cargo-DB and LLM connectors, the accounts/contacts models, folders) live in base-gtm: cookbooks import handles from there and never redefine them. The CRM connector is the one exception: it lives in crm-sync, so a cookbook that never touches a CRM does not inherit a credential it has no use for.
• Each cookbook's README states its resource graph, placeholders, and a "done when" check.

Cookbooks

The Requires column is the source of truth in cargo.scaffold.json and is enforced by npm run validate.

Folder	Outcome	Requires
base-gtm/	shared foundation: native accounts/contacts + credits connectors	—
crm-sync/	the CRM slot: the only cookbook that needs a CRM credential	base-gtm
tam-building/	Sales Nav company search, split past the 1,000 cap → accounts	base-gtm
list-building/	Sales Nav people search, split past the 2,500 cap → contacts	base-gtm
signal-based-tam/	signals over the TAM → living SAM + outreach play	base-gtm
inbound-flow/	form intake → verify → enrich → qualify → route + alert	base-gtm, crm-sync
contact-sourcing/	per SAM account: Cargo DB prospects → waterfall verify → CRM	base-gtm, crm-sync
routing-engine/	territories + capacity + territory-stamping play	base-gtm, crm-sync
account-scoring/	native scoring play + tier segments, criteria as code	base-gtm, crm-sync, gtm-knowledge-graph
auto-enrichment/	nightly waterfall freshness crons for accounts + contacts	base-gtm, crm-sync
crm-button/	run a Cargo tool from any CRM record, result written back	base-gtm, crm-sync
meeting-prep/	a briefing card in Slack before every intro call, deduped	base-gtm, crm-sync
pipeline-health/	agent flags at-risk deals with the rule + the fix, weekly digest	base-gtm
closed-won-multiplier/	every won deal → net-new lookalike accounts, deduped and traceable	base-gtm, pipeline-health
gtm-knowledge-graph/	the context repo as code + knowledge file + Q&A analyst	base-gtm
research-agent/	account → enriched brief via agent + context + tools	base-gtm, gtm-knowledge-graph
mcp-copilot/	the GTM stack behind one MCP endpoint	research-agent
ai-sdr/	sourcing + research + copywriter → reviewed outreach drafts	contact-sourcing, research-agent
plg-motion/	product events → PQL occurrence threshold → sales handoff	base-gtm, crm-sync
rep-cockpit/	hosted app: each rep's scored, routed book + drafts + next actions	account-scoring, routing-engine, ai-sdr

Already in Cargo: install these, do not build them

Two outcomes on the cookbook menu are prebuilt template plays in the Cargo template gallery, not CDK code. Writing them here would be reinventing something you can install in a minute.

Outcome	Where
Deduplication	Four template plays: companies and contacts, HubSpot and Salesforce. Search "dedu" in the template gallery, install the one matching your CRM and object, configure the matching keys and survivor rules, run.
Account hierarchy	Template play "List subsidiary and sisters companies of an account" (Salesforce + OpenAI + LinkedIn). Enriches the domain, works out whether the account is a parent, researches related entities, updates the account. Add a custom column mapping child / parent / standalone.

Zero configuration by default

base-gtm deploys with no credential and no environment variable: its accounts and contacts models are native (workspace-owned). A CRM key is needed only when you install crm-sync or a cookbook that requires it.

See CONTRIBUTING.md for local setup, checks, and how to add a cookbook.

Status: unverified. These cookbooks typecheck against the CDK and their scaffold graph validates, but they have not been cargo-ai cdk plan-ed or deployed against a live workspace yet. Treat the "Done when" section of each README as the acceptance test that still has to be run.