Updated step 3 to clarify payment method confirmation process and adjusted related instructions. Agent no longer needs to list out payment methods and input into spend request; should make the buying process faster
* update table styling for balances and sources commands
* test
* fix: rename Account name column to Source name for consistency
Committed-By-Agent: claude
- Handle nested `{ error: { message } }` error shape in auth resources
(both CLI and SDK implementations) by replacing the cast-only
`err?.error` interpolation with `extractOAuthErrorMessage`, which
does a runtime typeof check before descending into the object
- Fix `switch (err.error)` in `pollDeviceAuth` to switch on
`extractOAuthErrorCode(err)` so polling continues correctly when a
non-string error arrives (object would never match a string case)
- Fix `error_description` truthy guard to a nullish check so an empty
string from the server is preserved rather than falling back to the
error code
- Replace inline cast-only error extraction in payment-methods,
shipping-address, user-info, and web-bot-auth resources with
`extractErrorMessage` from base.ts, which already handles both
string and nested-object error shapes
- Replace `String(err)` in `useAsyncAction` with `JSON.stringify(err)`
so thrown plain objects produce readable output instead of
[object Object]
- Add tests covering all of the above
Committed-By-Agent: claude
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
The link-api PR (mint#2412595, LINK_AI_WALLET-365) added top-level
`card_brand` and `card_last4` to the RetrieveSpendRequest response so
integrators can identify a card for debugging without expanding the full
`card` object via `include=card`.
Add both fields to the SDK `SpendRequest` type and display them in the
interactive `retrieve` view when the full card is not expanded. The SDK
already passed these through to `--format json` output via its
pass-through cast; this closes the typing and interactive-display gap.
Docs (README, SKILL.md, CLAUDE.md) updated to match.
Committed-By-Agent: claude
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add a postinstall script that copies the SKILL.md into
~/.claude/skills/ and ~/.codex/skills/ so the skill is
available globally without being inside the repo. Updated
prepack to bundle the skills directory into dist/.
Committed-By-Agent: claude
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* Add MPP inspect command
* feat: rework mpp pay to handle full end-to-end 402 flow
mpp pay now probes the URL for a 402 challenge, parses the
WWW-Authenticate header to extract network_id and amount, creates a
spend request, yields the approval URL for the agent to present, and
completes payment with the SPT after approval.
- Remove mpp inspect (subsumed into pay flow)
- Add --context, --amount, --payment-method-id, --test flags
- Make --spend-request-id optional (backward compat)
- Agent mode yields approval URL immediately with _next hints
- Interactive mode handles full flow inline with polling
- Update SKILL.md and CLAUDE.md with new flow docs
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Committed-By-Agent: claude
* fix: update mpp pay tests to handle generator array output
The mpp pay command now uses an async generator (`async *run`) which
wraps yielded values in an array in JSON output mode. Update three
tests to extract [0] from the output array.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Committed-By-Agent: claude
* refactor: remove auto-generated context and step translation layer in mpp pay
Context is now required (min 100 chars) for the full MPP flow — agents must
provide a meaningful description. Also refactors onStep to emit typed Step
values directly, removing the string-to-enum mapping in the component.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Committed-By-Agent: claude
* revert version bump in SKILL.md
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Committed-By-Agent: claude
* fix: remove extra blank line to pass biome format check
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Committed-By-Agent: claude
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(mpp): support stripe session challenges alongside charge
Add intent="session" support to the MPP flow. Session challenges use an
open/grantedToken credential shape instead of the bare SPT payload.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Committed-By-Agent: claude
* chore: retrigger CI
Committed-By-Agent: claude
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* docs(create-payment-credential): make Link Pay Token flow autonomous-safe
The Link Pay Token section covered token retrieval and injection but
omitted the pieces that make the flow complete unattended, so an agent
following it would fail detection/injection and retry in a loop:
- the "I am an AI agent" checkbox that gates the hidden
input[name="link_pay_token"] (and must be set via the native setter,
not .click());
- that the checkbox and input live in the Stripe accessory iframe;
- the ~5-minute token TTL (retrieve just before injecting);
- the exchange/lookup success signal to wait on;
- a bounded stop/fallback instead of endless retries.
Changes (docs only, merchant-agnostic; no internal flags/accounts):
- Step 2 detection now keys on the checkbox, not only the initially
absent token input.
- Step 5 Link Pay Token: fresh context -> reveal via checkbox -> retrieve
token -> inject -> wait for go-signal -> pay, plus "stop, don't loop".
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Committed-By-Agent: claude
* docs(create-payment-credential): gate LPT flow on the input, not the component
Review feedback (raubrey): lean on the AiAgentPaymentSteering block as the
anchor and instruction source instead of duplicating mechanics, and account
for surfaces that render the block but disable the inline token input.
The same component renders on Payment Elements (input on by default) and on
hosted Checkout (input gated by checkout_hosted_link_pay_token_enabled, off
today). So "found the component" cannot mean "do the token flow": on hosted
Checkout the checkbox and instructions show with no input, which would loop.
- Step 2 detection now confirms the token path by checking whether
input[name="link_pay_token"] appears after ticking the checkbox; if it does
not, follow the block's instructions and fall back to card.
- Step 5 delegates the flow narrative to the block's rendered steps, reveals
the checkbox with a DOM-level click() (matching the e2e deploy spec) instead
of a native-setter dance, and keeps the bounded stop/fallback so an
input-disabled surface never loops.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Committed-By-Agent: claude
* update skill.md
* LPT to card fallback can reuse the same spend request
* udpate payment element and browser session detail
* drop references to elements
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat: surface verification_url in spend-request error output
When the API returns additional_verification_required, surface the
verification_url so users know where to complete stepup verification.
Also fixes interactive mode hanging on error by calling useApp().exit().
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Committed-By-Agent: claude
* fix: resolve biome lint and format errors in spend-request commands
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Committed-By-Agent: claude
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
github-actions[bot]