Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Expert patterns for Plaid API integration including Link token flows, transactions sync, identity verification, Auth for ACH, balance checks, webhook handling, and fintech compliance best practices. Use when: plaid, bank account linking, bank connection, ach, account aggregation.
.claude/skills/davila7-plaid-fintech/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-03 | ✓→✓ | = Same ✓ | -29% | 0% |
Create a link_token for Plaid Link, exchange public_token for access_token. Link tokens are short-lived, one-time use. Access tokens don't expire but may need updating when users change passwords.
Use /transactions/sync for incremental transaction updates. More efficient than /transactions/get. Handle webhooks for real-time updates instead of polling.
Handle ITEM_LOGIN_REQUIRED errors by putting users through Link update mode. Listen for PENDING_DISCONNECT webhook to proactively prompt users.
| Issue | Severity | Solution | |-------|----------|----------| | Issue | critical | See docs | | Issue | high | See docs | | Issue | high | See docs | | Issue | high | See docs | | Issue | medium | See docs | | Issue | medium | See docs | | Issue | medium | See docs | | Issue | medium | See docs |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 19,093 | 14,086 | -26% | 1 | 1 | 0% | 3,078 | 2,934 | -5% | 0 | 0 | — |
case-02 | fail→pass | 17,466 | 13,811 | -21% | 1 | 1 | 0% | 2,717 | 2,731 | +1% | 0 | 0 | — |
case-03 | pass→pass | 17,891 | 10,437 | -42% | 1 | 1 | 0% | 3,174 | 2,250 | -29% | 0 | 0 | — |
case-04 | pass→pass | 19,881 | 13,590 | -32% | 1 | 1 | 0% | 2,989 | 2,558 | -14% | 0 | 0 | — |
case-05 | pass→pass | 14,528 | 8,020 | -45% | 1 | 1 | 0% | 2,301 | 1,449 | -37% | 0 | 0 | — |
case-06 | pass→pass | 12,807 | 10,645 | -17% | 1 | 1 | 0% | 2,150 | 1,880 | -13% | 0 | 0 | — |
case-07 | pass→pass | 15,963 | 14,753 | -8% | 1 | 1 | 0% | 2,590 | 2,237 | -14% | 0 | 0 | — |
case-08 | pass→pass | 9,024 | 6,383 | -29% | 1 | 1 | 0% | 1,465 | 1,426 | -3% | 0 | 0 | — |
case-09 | fail→pass | 12,538 | 4,350 | -65% | 1 | 1 | 0% | 2,053 | 1,051 | -49% | 0 | 0 | — |
case-10 | fail→pass | 22,825 | 7,441 | -67% | 1 | 1 | 0% | 2,081 | 1,506 | -28% | 0 | 0 | — |
case-11 | pass→pass | 16,167 | 12,269 | -24% | 1 | 1 | 0% | 2,345 | 2,224 | -5% | 0 | 0 | — |
case-12 | pass→pass | 15,558 | 13,054 | -16% | 1 | 1 | 0% | 2,698 | 2,391 | -11% | 0 | 0 | — |
case-13 | pass→pass | 11,601 | 7,282 | -37% | 1 | 1 | 0% | 1,906 | 1,456 | -24% | 0 | 0 | — |
case-14 | pass→pass | 15,239 | 12,507 | -18% | 1 | 1 | 0% | 2,584 | 2,308 | -11% | 0 | 0 | — |
case-15 | pass→pass | 5,463 | 3,819 | -30% | 1 | 1 | 0% | 1,080 | 853 | -21% | 0 | 0 | — |
case-16 | pass→pass | 12,192 | 6,803 | -44% | 1 | 1 | 0% | 2,045 | 1,411 | -31% | 0 | 0 | — |
case-17 | pass→pass | 11,857 | 8,682 | -27% | 1 | 1 | 0% | 2,058 | 1,771 | -14% | 0 | 0 | — |
case-18 | pass→pass | 19,066 | 14,900 | -22% | 1 | 1 | 0% | 3,279 | 2,819 | -14% | 0 | 0 | — |
case-19 | pass→pass | 10,908 | 8,390 | -23% | 1 | 1 | 0% | 1,699 | 1,615 | -5% | 0 | 0 | — |
case-20 | pass→pass | 9,365 | 11,133 | +19% | 1 | 1 | 0% | 1,830 | 1,423 | -22% | 0 | 0 | — |
case-21 | pass→pass | 13,815 | 12,434 | -10% | 1 | 1 | 0% | 2,415 | 2,259 | -6% | 0 | 0 | — |
case-22 | pass→pass | 6,437 | 5,442 | -15% | 1 | 1 | 0% | 1,238 | 1,403 | +13% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.