the corpus is what makes rightstack useful. the more accurate the tool and workflow records, the better the recommendations. contributions are welcome.
open an issue — no code needed. use the request a tool or request a workflow templates and describe what's missing.
open a pr — add a json file to data/tools/ or data/workflows/ directly.
create data/tools/<tool-id>.json. use an existing file as reference, data/tools/privy.json is a good start.
required fields:
{
"id": "your-tool-id",
"name": "Display Name",
"aliases": [],
"category": "wallet-infrastructure",
"description": "one sentence: what it does and when to use it.",
"ecosystem_fit": [
{ "ecosystem": "base", "strength": "dominant" }
],
"trust_state": "production-grade",
"anti_patterns": [],
"retrieval_tags": [],
"source": {
"derived_from": "your name or source",
"last_verified": "YYYY-MM-DD"
},
"updated_at": "YYYY-MM-DD"
}trust states: production-grade / beta / experimental / deprecated
ecosystem strength: dominant / strong / present / minimal
create data/workflows/<workflow-id>.json. use data/workflows/base-consumer-app.json as reference.
a workflow is a build goal with phases. each phase maps to a tool category and recommends specific tools. this is what powers rightstack recommend.
required fields: id, name, goal, ecosystems, phases, trust_state, tradeoffs, anti_patterns
each phase needs: id, role, required, primary_tools, alternative_tools
cd apps/cli
npm run build
node bin/rightstack.js benchmark invariantsall invariants must pass.
- specific over generic. describe the tool in the context of what rightstack covers: consumer crypto apps, ai agents, onchain apps
- honest trust states. beta is beta. deprecated is deprecated
- real anti-patterns. what do devs actually get wrong with this tool
- accurate ecosystem fit. only mark dominant if it genuinely is
- tools outside scope: validator infra, consensus engineering, general-purpose backend libraries
- records without a trust state
- records copied from marketing docs without independent assessment