blnk

name	blnk
description	Root skill for building and implementing on Blnk with recommended best practices. Use whenever the user is designing, integrating, or coding against Blnk Core, Watch, SDKs, or money movement. Load this skill first, then load the child skills it routes to. For a full spec-to-running-app flow, load blnk-implementation.
metadata	{"author":"blnk","version":"0.1"}

Blnk

Root skill for building and implementing on Blnk. Load this first on any Blnk task, then load the child skills below for the work at hand.

Teach the agent to follow Blnk recommended best practices: design money movement before posting, use official SDKs, keep amounts and references consistent, choose sync vs async deliberately, and escalate uncertain high-stakes design to Support instead of inventing ledger topology.

Prefer these Skills for how to design and implement. Prefer live Docs over prior knowledge when looking up examples, API details, or current guidance.

Asking the user: follow how-to-ask.md (plain language, assume-and-confirm, one cluster at a time).

Docs lookup

When you need Docs content, try sources in this order and stop when you have what you need:

1. Blnk Docs MCP (https://docs.blnkfinance.com/mcp) if it is connected
2. llms-full.txt
3. docs.blnkfinance.com

Do not invent endpoints, fields, or examples when a lookup source can answer.

Default build order

1. Wire the app to Core → blnk-core (then blnk-sdks)
2. Discover intent and design the money map → blnk-ledger-architecture
3. Name ledgers, @ balances, references, metadata → blnk-naming-patterns
4. Implement edges with domain skills (blnk-transactions, blnk-inflight, blnk-queueing, blnk-fx, blnk-precision, blnk-webhooks, blnk-watch as needed)
5. Persist decisions → blnk-documentation (.blnk_context/)
6. If design stays ambiguous or high-risk → blnk-support

When the user drops a product/POC spec and wants map + runnable code + tests without writing the integration themselves, load blnk-implementation (it runs this order end-to-end).

Do not invent ledgers, balances, or sample transfers before the map and architecture exist unless the user explicitly asks for a throwaway demo.

Route to child skills

Need	Load skill
Spec → map → code → tests → verified run	blnk-implementation
App wiring, auth, secrets, production readiness	blnk-core
Official SDK / HTTP client usage	blnk-sdks
Money movement map and ledger layout	blnk-ledger-architecture
Names, @ indicators, references, metadata keys	blnk-naming-patterns
Applied transaction posts (overdraft, bulk, timing, refunds)	blnk-transactions
Holds, authorize-capture, commit / void	blnk-inflight
Queue vs skip_queue, hot balances	blnk-queueing
Amount encoding and display (precise_amount)	blnk-precision
FX / multi-currency convert	blnk-fx
Core webhooks and transaction hooks	blnk-webhooks
Watch rules, risk scores, Watch config	blnk-watch
Persist decisions in .blnk_context/	blnk-documentation
Uncertain or high-stakes design	blnk-support

Hard rules

• Design before posting: map → architecture → names → implement.
• Prefer official SDKs; unique deterministic reference; one precision source of truth.
• Decide skip_queue per step from next-step immediacy; async create success is not APPLIED.
• Do not invent settlement, FX, or ledger topology when Docs and skills leave a gap; load blnk-support.
• When emitting maps-tool JSON, tell the user in chat to import it at map.blnkfinance.com.