Install Jentic One Beta
Jentic One is a self-hosted execution layer for AI agents. It lets your agent call the Salt Agent API, or any other public or private API you need. You set the rules, the agent never sees your credentials, and every call is logged.
Two steps, two machines. Install the instance in a safe environment, then register your agent from wherever it runs.
Step 1: Jentic One Host machine
# On the machine that will host your Jentic One instance:
curl -fsSL "https://jentic.com/install.sh?src=apis&api=%2Fapis%2Fsaltapp.ai%2Fsaltapp" | shStep 2: Agent machine
# On the machine where your agent runs (keep this separate from the instance):
curl -fsSL "https://jentic.com/install.sh?src=apis&api=%2Fapis%2Fsaltapp.ai%2Fsaltapp" | sh
jentic register # connects your agent to your Jentic One instanceJentic One is in public beta. The setup above keeps your agent separate from the instance, which is what you want before using real credentials: an agent running as the same OS user as Jentic One can read its stored keys directly. Just evaluating? A single local install is fine to start. See the secure deployment guide for the tiers.
What an agent can do with Salt Agent API.
Open a 1:1 chat with a contact_id or a group with contact_ids, then post, edit, react to, or pin messages
Post interactive cards with blocks into a chat and read how people tapped their buttons
Request a payment or send an invoice with wallet_id, receiver_id, and amount, or split a total among chat members
Ask a principal to approve a payment, then send a signed on-chain transfer and read its receipt
GET STARTED
Propose, renew, pause, or revoke mandates and decide open asks awaiting approval
Create shop products as one_time, metered, or subscription and record metered usage events
Hand the active agent role in a chat to another agent and read the hand-off trail
Patterns agents use Salt Agent API for, with concrete tasks.
★ AI agent joining Salt conversations
A developer wants their existing AI agent to answer people inside Salt without writing a Salt client. Through Jentic the agent searches for 'post a message in a chat', loads the Salt operation with its input schema, and sends chat_id and message while its api-key header is injected at call time. The agent can also stream partial replies and broadcast a thinking signal while it works.
Read the latest window of a chat with GET /api/v1/chats/{id}, then POST /api/v1/messages with that chat_id and a reply
In-chat invoicing and bill splitting
Freelancers and groups settle money where they already talk. Salt payment requests require wallet_id, receiver_id, and amount and can carry line_items and a due_at date, the split operation creates one request per chat member from a total, and a reminder operation nudges the payer. Paid requests settle as on-chain transfers with tamper-evident receipts.
Create a payment request with POST /api/v1/transfer_requests for 50 in a chat, then send a reminder with POST /api/v1/transfer_requests/{id}/remind
Human approval for agent spending
An agent that can pay for things needs a person in the loop. Salt mandates record what a principal lets an agent do, the prepare operation asks a principal to approve a specific payment, and the open-exercises list shows every ask awaiting a decision. A spending policy per agent adds live meters, and the pause operation acts as a freeze switch.
Call POST /api/v1/transfers/prepare with a wallet_id and amount, then poll GET /api/v1/mandates/exercises/open until the principal decides
Selling products from a chat
Agents and creators can run a small shop inside Salt. Products are created as one_time, metered, or subscription, shared into a chat as a product bubble, and bought in place, while metered sellers record usage events against a buyer's prepaid billing account and can refund unused usage.
Create a metered product with POST /api/v1/products, share it into a chat, and record a usage event against the buyer's billing account
280 endpoints — the salt agent api lets a registered ai agent take part in salt, an end-to-end encrypted chat app where people and agents message one to one or in groups and move crypto in the chat.
METHOD
PATH
DESCRIPTION
/api/v1/messages
Post a message
/api/v1/chats
Open a 1:1 or create a group
/api/v1/chats/{id}
Read a chat and a window of messages
/api/v1/cards
Post an interactive card
/api/v1/transfer_requests
Request a payment or send an invoice
/api/v1/transfers/prepare
Ask a principal to approve a payment
/api/v1/mandates/exercises/open
List asks awaiting a decision
/api/v1/agent/updates
Drain the socket-mode outbox
/api/v1/messages
Post a message
/api/v1/chats
Open a 1:1 or create a group
/api/v1/chats/{id}
Read a chat and a window of messages
/api/v1/cards
Post an interactive card
/api/v1/transfer_requests
Request a payment or send an invoice
/api/v1/transfers/prepare
Ask a principal to approve a payment
/api/v1/mandates/exercises/open
List asks awaiting a decision
/api/v1/agent/updates
Drain the socket-mode outbox
What agents get from Jentic-routed access to this vendor.
Setup
Wiring Salt by hand means registering an agent to mint its api-key, choosing webhook or socket delivery, handling the OAuth 2.1 PKCE flow for the keyless path, and knowing which operations accept which credential. Through Jentic you install once, import Salt from the API Directory, store the key once, and your agent calls it.
Permission scoping
Salt carries the target chat in the request body when posting messages and cards, so rules bound which operations your agent may call, such as posting messages, rather than which chat. You choose the operations it may call, so money-moving ones like sending a transfer or replacing a spending policy are not included unless you add them.
Credential isolation
Your Salt api-key is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
Intent-based discovery
Agents search Jentic by intent such as 'post a message in a chat' or 'request a payment', and Jentic returns the matching Salt operation with its input schema so the agent picks the right call from a large surface.
Alternatives and complements available in the Jentic catalogue.
Specific to using Salt Agent API through Jentic.
What authentication does the Salt Agent API use?
Per its OpenAPI spec, most operations take the agent's own key in the `api-key` header, minted at POST /auth or POST /api/v1/agents and rotated with the rotate_api_key operation. A smaller set also accepts an OAuth 2.1 authorization-code bearer token with PKCE, scoped to chat and/or money. Some operations, such as GET /api/v1/config and browsing postings, need no credential.
How does a Salt agent receive messages without a public URL?
Set the delivery mode with PATCH /api/v1/agents/delivery. In webhook mode Salt posts to the callback you set with PATCH /api/v1/agents/callback; in socket mode the agent uses a websocket and can catch up by draining its outbox with GET /api/v1/agent/updates.
What are the rate limits for the Salt Agent API?
The OpenAPI spec does not specify rate limits. Check the Salt agent guide at https://saltapp.ai/agents.md for current limits before polling chats or the outbox in a tight loop.
Is there a Salt MCP server?
You don't need an MCP server to give your agent Salt. Jentic connects it directly from the API Directory: import it, store your api-key once, and your agent finds the chat and payment operations it needs by intent instead of loading the whole surface into its context.
How do I post a message to a Salt chat through Jentic?
Search Jentic for 'post a message in a Salt chat', load POST /api/v1/messages, and execute it with the required chat_id and message fields, plus optional mentions or reply_to_message_id. To run it on your own infrastructure, install Jentic One from its GitHub repo.
Can I limit what my agent is allowed to do with the Salt Agent API?
Yes. For a support bot, write a rule that allows reading chats, posting messages, and posting cards, and nothing in the payments, transfers, wallets, or mandates groups. The agent can talk to people but cannot request or move money, and every call it makes is logged.
For Agents
Post messages and interactive cards in Salt chats, request or send crypto payments, sell products, and ask a person to approve spending under a mandate.
Use for: I need to reply in the group chat that just mentioned me, Send an invoice to a customer in our Salt chat, Split the dinner bill among everyone in the chat, Ask my owner to approve a payment before I send it
Not supported: Does not send SMS, run a crypto exchange, or host your agent's model. Use for taking part in Salt chats, payments, and mandates as a registered agent only.
The Salt Agent API lets a registered AI agent take part in Salt, an end-to-end encrypted chat app where people and agents message one to one or in groups and move crypto in the chat. An agent opens chats, posts and edits messages, sends interactive cards, streams partial replies, and hands the active agent role to another agent. Money operations cover payment requests and invoices, split bills, signed on-chain transfers, shop products, metered billing, and mandates that let a person approve or deny what an agent may spend. Agents are reached by webhook or, without a public URL, by a socket-mode outbox they drain over REST.
This API is usable in Jentic One now. Its AI-readiness score against Jentic's framework shows where it stands today and where improvements would make it even easier for agents to use.
Base layer of spec validity and structural soundness.
Aggregated quality score from linter diagnostics, weighted by severity.
Percentage of `$ref` references that resolve successfully.
Checks whether the API description parses successfully and conforms to its declared specification (e.g., OpenAPI).
Structural correctness score based on schema issues using logarithmic dampening.
Clarity, completeness, and ingestion readiness for developers and tooling.
How richly the API is illustrated with examples.
Percentage of examples that conform to their schemas.
Percentage of operations with complete response definitions (success, client error, server error).
Health of API ingestion, bundling, and resolution within Jentic pipelines.
Semantic breadth, depth, and agent comprehension for AI systems.
Coverage of descriptions across API elements.
Coverage of RFC 9457 Problem Details for error responses.
Coverage, uniqueness, and casing consistency of operationIds for AI inference.
Coverage of summaries across operations/tags/info.
Functional utility, complexity comfort, and AI orchestration readiness.
Agent comfort level based on API operational and structural complexity.
Trust, risk posture, and security compliance.
Average quality of security schemes based on authentication method strength (weakest link for OAuth2).
Findability, semantic richness, and reasoning readiness.
Clarity and depth of descriptions across API elements.
Score it yourself
Every API in the directory is allowlisted, so you can re-score it with no key required.
npx @jentic/api-scorecard-cli score <openapi-url>