supplement_gradingLABSCO SUMMARY
Every skill carries its price, and the free allowance is on the one you would call most.
Prices run from a tenth of a cent for a biomarker sync to eight tenths for a lab panel, published per skill rather than hidden in a plan, which makes the cost of an agent loop predictable before you write it. The dependency worth understanding first is the phone: the account and the wearable data both come from the iOS app, so an agent-only setup reaches the chat, lab and food skills but not the CGM ones. The shipped routing file is a nice touch — it lets an agent choose the right skill without you enumerating them in a prompt.
What it is
An interface to Sally, A1C's metabolic-health service. Seven skills are exposed as tools, each priced per call against a wallet balance, reading the data the A1C Insights iPhone app collects through HealthKit. The same skills are reachable over REST for anything that cannot speak MCP.
What you get
- 64+ live biomarkers pulled from wearables and CGM into the agent's context — 10 calls free each month, then $0.001 each (health_sync).
- A full CGM snapshot at $0.005 a call: time in range, variability and postprandial curves (metabolic_overview).
- Lab panels parsed and interpreted at $0.008 a call, with risk flags and reference ranges attached (analyze_lab_result).
- Meals graded by macros and predicted glucose response at $0.004 a call, with patterns surfaced across entries (food_journal).
- Morning, afternoon and evening readouts assembled from sleep, vitals and activity at $0.003 a call (health_insights).
- A supplement stack graded for interactions, evidence quality and gaps at $0.008 a call (supplement_grading).
- Open questions answered across preventive health and Traditional Chinese Medicine, evidence-graded, at $0.003 a call (chat_with_sally).
- A routing file you can drop into an agent's system prompt so it picks the right skill on its own, with decision tables, chaining patterns and anti-patterns.
Requirements
An A1C account, which is created in the A1C Insights iPhone app — the account itself cannot be made anywhere else. Then an API key minted at console.a1c.io, in the sk-sally- form, shown once. Each key is bound to one account; several keys can share one wallet balance, and per-key usage stays auditable. Three skills read CGM and wearable data collected through HealthKit, so they depend on the phone app having that data; the chat, lab and food-journal skills work without any wearable. Calls draw on a funded wallet at the per-call prices above.
Setup effort
Follow the published steps
USE CASES
Getting your own numbers into a conversation
Pull the current biomarkers from your wearables and CGM rather than describing them from memory.
Making sense of a lab report you just received
The panel is parsed with reference ranges and risk flags attached to each value.
Understanding how a meal affected you
The meal is graded by macros and predicted glucose response, with patterns across entries.
Reviewing what you are already taking
The stack is graded for interactions, evidence quality and gaps rather than item by item.
Reading how the day went
Sleep, vitals and activity assembled into a readout instead of four separate app screens.
7 tools
Listed in project docs. Your client may show a different list.
health_syncchat_with_sallyanalyze_lab_resulthealth_insightsfood_journalmetabolic_overviewBEFORE YOU INSTALL
What to know first
Tools you get
7 tools named in the project's documentation.
INSTALLATION
Connect your client
Server setup
Install the A1C Insights iOS app from the App Store (https://apps.apple.com/id/app/a1c-insights/id6748399956) and sign up — the account can only be created there
Sign in at https://console.a1c.io with that same account, open API Keys → Create new key → Generate, and copy the sk-sally-… key (shown only once)
Top up the wallet at https://console.a1c.io/billing (per-call pricing, no subscription)
Smoke-test the key: curl -sS https://sally.a1c.io/v1/call -H "Authorization: Bearer sk-sally-…" -H "Content-Type: application/json" -d '{"skill":"health_sync","input":{}}' | jq .ok
Add the MCP config from protocols/mcp.md (or the per-agent guide in agents/) and restart the agent