cargo-manifest
llms.txt12 starsChanged 18 days agoMITView on GitHub
A free, open-source template for running go-to-market the way engineers run software: everything a company knows, everything it is trying to do, and everything that runs in production, in one repository the team owns and AI agents can operate. Built by Cargo (https://getcargo.io). MIT licensed. Manifest structures a GTM organization as a codebase. Knowledge lives in markdown with a confidence ladder, the live revenue engine is declared as code and deployed to a Cargo workspace, agent procedures are versioned skills,…
In this repository8 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .cursor/rules/manifest-guardrails.mdcCursor rule
- cadenceSkill
- capture-feedbackSkill
- context-hygieneSkill
- new-executionSkill
- new-motionSkill
The file
Files
# Manifest > A free, open-source template for running go-to-market the way engineers run software: everything a company knows, everything it is trying to do, and everything that runs in production, in one repository the team owns and AI agents can operate. Built by Cargo (https://getcargo.io). MIT licensed. Manifest structures a GTM organization as a codebase. Knowledge lives in markdown with a confidence ladder, the live revenue engine is declared as code and deployed to a Cargo workspace, agent procedures are versioned skills, and every run is archived with the outcome it produced. Agents read the repo before acting, so they reason from the company's own knowledge instead of generic priors. ## Layers - [AGENTS.md](AGENTS.md): the agent contract. Read order, layer rules, conventions. Every agent reads this first. - [plan/](plan/README.md): the destination. The one goal, the strategy, and the outcomes that must become true. Each outcome has a metric defined well enough to calculate, a baseline, a target, and names what it becomes in infra/. An outcome an agent owns also sets its weekly signal, guardrails, experiment budget, decision rights and escalation point. - [context/](context/README.md): what the company knows. Eleven domains (global, icp, persona, jtbd, alternative, objection, proof, signal, motion, client, insight) as markdown with YAML frontmatter and a knowledge graph. - [initiatives/](initiatives/README.md): bounded efforts with owners, deadlines, and success criteria checkable as true or false. - [cadence/](cadence/README.md): the operating rhythm. Weekly plans, daily logs, and a carryover file that keeps dropped balls visible. - [infra/](infra/README.md): what runs in production. A Cargo CDK project declaring connectors, models, segments, tools, agents, plays, MCP servers, apps, and workers. Ships flat and near-empty; cookbooks add the structure. Previewed as a diff on every PR, deployed on merge. - [evals/](evals/promptfooconfig.yaml): promptfoo regression suites for agent prompts, run with `npm run eval`. - [outputs/](outputs/README.md): the append-only archive of what happened, each entry tagged with the outcome it produced. ## Conventions - Confidence ladder: proof, signal, and insight files carry `confidence: hypothesis | validated | proven`. Outreach motions may cite only validated or proven knowledge; the linter enforces it. - Outreach motions declare a hypothesis (segment, angle, trigger, expected reply rate) before any copy exists. Enrichment, scoring, routing, and signal motions are not campaigns and are exempt. - A claim enters context after it shows up twice, independently. Once is an anecdote. - Outputs entries carry an `outcome:` field, so motions are ranked by results rather than opinions. - One-off runs (enroll a list into a play, backfill a field) are declared as code in an outputs entry, previewed on the pull request, and run exactly once by the merge. - Absolute dates everywhere. Outputs are append-only. Major context rewrites archive the prior version. ## Cargo - [Cargo](https://getcargo.io): the platform the infra layer deploys to. Unifies GTM data into clean account and contact tables, then runs enrichment, scoring, routing, a context graph, and AI agents on top. - Getting a workspace takes two terminal commands and no browser: `npx @cargo-ai/cli login --email you@company.com` emails a one-time code and exits, then the same command with `--code <code>` creates the account and a workspace. Where there is no terminal to prompt at (agent shells, sandboxes), that send-then-exit behaviour is what makes it work. - [Cargo docs](https://docs.getcargo.ai): CLI, CDK, agents, connectors. - [Cookbooks](https://github.com/getcargohq/cargo-cookbooks): prebuilt recipes (deduplication, contact sourcing, routing, meeting prep) that scaffold straight into infra/.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

