Skip to content
Agentmart
Submit a tool

aba-payway-agent

Use when operating the agentic PayWay CLI with provider modes, the 14 tools (incl.

Skill by Ruzaid-aman

IndexedMIT
Write a review

Chalk Labs

Need help deploying aba-payway-agent?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/ruzaid-aman-aba-payway-agent

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Use when operating the agentic PayWay CLI with provider modes, the 14 tools (incl. query_knowledge), risk gates, the execution ledger, sessions, local journal queries, knowledge-base search, and secret redaction.

Use cases: security, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/aba-payway-agent

What's inside

Skill name
aba-payway-agent
Description
Use when operating the agentic PayWay CLI with provider modes, the 14 tools (incl. query_knowledge), risk gates, the execution ledger, sessions, local journal queries, knowledge-base search, and secret redaction.
License
MIT
Metadata
version: 1.4.1
Instructions
281 lines in SKILL.md · view on GitHub

Before installing

  • Instructions read files in your home folder (~/.ssh, ~/.aws…)

Files (1)

  • SKILL.md

SKILL.md (first lines)

# ABA PayWay Agent (Agentic CLI)

The agentic CLI lets a supported provider propose and run PayWay actions through a
risk-gated pipeline. It is the safest way to let an LLM drive PayWay: **no
side-effecting call runs unless the plan validates, is authorized, and is
materialized through the execution ledger.**

## Quick Start

```sh
# 1. Configure the provider (API key stays in the environment, never stored).
#    Presets: opencode (free) | openai | openrouter | nvidia | custom
export PAYWAY_AGENT_API_KEY=sk-...
payway-sdk agent setup --provider opencode --model deepseek-v4-flash-free --capability-mode strict-json-plan \
  --max-tokens 8192 --temperature 0.2
# For NVIDIA thinking models add: --extra-body '{"chat_template_kwargs":{"enable_thinking":false}}'

# 2. First run must acknowledge the privacy notice, then ask.
payway-sdk ask "Generate a $3 online QR for sandbox" --yolo
```

The provider config shape (stored as plaintext, no secrets; managed by
`payway-sdk agent setup` — you normally never write it by hand) is:

```ts
// ProviderConfigV1 lives in the SDK source (src/agent/contracts.ts).
// The package exports only the root entrypoint ("."), so import the
// PayWay client from 'aba-payway-ts' and manage agent config via the CLI:
//   payway-sdk agent setup --provider opencode --model deepseek-v4-flash-free \
//     --capability-mode strict-json-plan
const config = {
  version: 'agent-config/v1',
   provider: 'opencode',               // 'opencode' | 'openai' | 'openrouter' | 'nvidia' | 'custom'
   model: 'deepseek-v4-flash-free',    // rotate as the provider's catalog changes — a dead model fails with 401 "not supported"
  capabilityMode: 'strict-json-plan', // 'native-tools' | 'strict-json-plan'
  privacyAcknowledgedAt: new Date().toISOString(),
};
```

## Provider capability modes

- **`native-tools`** — the provider emits tool calls directly (function calling).
  Every call is validated against the tool schemas before execution.
- **`strict-json-plan`** — the provider returns a single JSON plan of `actions`;
  the orchestrator normalizes, authorizes, and executes them one by one. This is
  the default and the most auditable mode.

Both modes funnel through the same safety pipeline; only the plan source differs.

## The 14 tools

| # | Tool | Create? | Purpose / key inputs |
|---|---|---|---|
| 1 | `generate_online_qr` | yes | Online QR via PayWay API. `amount`, `currency`, `callbackUrl`, `transactionId?` |
| 2 | `generate_offline_khqr` | yes | Local-only ABA KHQR. `amount?`, `currency`, `merchantRef` |
| 3 | `create_checkout_payload` | yes | LOCAL signed checkout payload (no network). `amount`, `currency`, `transactionId?` |
| 4 | `create_checkout_purchase` | yes | NETWORK checkout request to PayWay. `amount`, `currency`, `transactionId?` |
| 5 | `create_payment_link` | yes | Shareable link (needs RSA). `title`, `amount`, `currency`, `merchantRefNo`, `returnUrl`, `payout?` |
| 6 | `get_payment_link_details` | no | Read-only payment-link lookup by `paymentLinkId` (the opaque Link ID from create) |
| 7 | `check_transaction` | no | Read-only status lookup by `transactionId` |
| 8 | `check_transaction_by_merchant_ref` | no | Read-only lookup by `merchantRef` |
| 9 | `poll_transaction` | no | Read-only repeated lookup (sandbox verification) by `transactionId` |
| 10 | `query_journal` | no | Read-only local journal query: `timeline`, `stats`, `reconcile`, or `anomalies` |
| 11 | `save_artifact` | no | Persist a produced artifact to `./payway-output` |
| 12 | `open_artifact` | no | Open a saved artifact |
| 13 | `copy_to_clipboard` | no | Copy an artifact value to the clipboard |
| 14 | `query_knowledge` | no | Read-only offline knowledge base: `query: search|read` over the curated PayWay corpus (`payway-sdk docs`) — pull integration guidance, error hints, and sandbox-verified lessons into a plan |

> `create_checkout_payload` builds a **local** signed payload; `create_checkout_purchase`
> performs the actual **network** request. Treat them as distinct tools.

## Resolved context (no ambient override)

The effective PayWay context is resolved **once** from the selected CLI profile
(`--profile` / `PAYWAY_PROFILE` / saved default / active profile). A selected
profile's credentials are authoritative; **stale ambient `PAYWAY_MERCHANT_ID` /
`PAYWAY_API_KEY` values MUST NOT override it**, and a provider plan can never
change the selected profile or environment. The context is what the executor
uses — the provider only receives a scrubbed, secret-free summary.

## Risk gates

| Invocation | Sandbox create | Production create |
|---|---|---|
| `ask ... --approve` | allowed | allowed (authorizes both) |
| `ask ... --yolo` | allowed (sandbox only) | **refused** — `--approve` required |
| `ask ...` in a TTY | interactive confirmation | interactive confirmation |
| `ask ...` in non-TTY, no flag | `needs_confirmation` — **no create** | `needs_confirmation` — **no create** |

Read-only tools (`get_payment_link_details`, `check_transaction`,
`check_transaction_by_merchant_ref`, `poll_transaction`, and `query_journal`)
never require approval. Plans containing only read-only
actions run without a flag. A resumed session restores **no** stored approval —
it returns `needs_confirmation` and must be re-confirmed.

## Execution ledger

Every create action passes through versioned ledger records with a strict,
non-replayable lifecycle:

```
planned -> confirmed -> submitted -> succeeded | failed | outcome_unknown
```

- `planned`: created when the plan is accepted (no PayWay call yet).
- `confirmed`: a `transactionId` is generated and bound to the materialized plan.
- `submitted`: the single SDK call runs.
- terminal: `succeeded`, `failed`, or `outcome_unknown` (network/timeout — verify before retrying).

Ledger records are **never replayed**. A crash between `confirmed` and `submitted`
executes zero PayWay creates.

## Local utilities

- `payway-sdk agent doctor` — prints the capability matrix (provider connectivity,
  context, online QR, offline KHQR, checkout, payment-link RSA, artifact/session storage).
  Every non-ready row includes a `→` fix hint pointing at the command that resolves it.
- `payway-sdk onboard` — **guided interactive setup**. Scans state, shows a
  configured/missing checklist, then runs the needed stages in order: inference provider

From Ruzaid-aman/ABA-Payway-SDK-unofficial, MIT.

More from Ruzaid-aman/ABA-Payway-SDK-unofficial

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Oct 1, 2026 (1 month ago)

From GitHub (fetched Oct 24, 2026).

Details

Version
1.4.1
License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from Ruzaid-aman

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get aba-payway-agent running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying aba-payway-agent?

CallGet help
Cite this page

aba-payway-agent skill. Agent Mart. https://agentmart.dev/skills/ruzaid-aman-aba-payway-agent (last updated 2026-10-24)