Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →End-to-end onboarding of a new entity, customer, or account into TRES Finance. Orchestrates 8 steps in order: wallet upload, data collection (commit), balance validation, reconciliation, cost basis, export unidentified addresses, import contacts, and rollup rules. Trigger ONLY when the user explicitly wants to onboard a new entity, customer, or account — e.g. "onboard a new entity", "onboard new customer", "start onboarding for [name]", "set up a new account in TRES", "run the full onboarding fl
.claude/skills/anthropics-tres-onboarding/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 17 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 60% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 159% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 149% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 146% | 0% |
Orchestrate the complete onboarding of a new entity/customer/account into TRES Finance. This skill does not contain its own low-level logic — instead, it drives the user through 8 sequential steps, invoking the dedicated skill for each one. Think of it as a conductor that keeps the onboarding moving forward while each sub-skill handles the details.
No visualizations for input gathering. Do NOT use show_widget, HTML widgets, interactive forms, AskUserQuestion widgets, or any visualization tool when collecting initial context from the user or presenting the onboarding roadmap. Ask all questions and present the roadmap as plain text in the chat. This keeps the flow fast and responsive — building a widget just to ask for a name and wallet list adds unnecessary latency. Visualizations are fine later when individual sub-skills need them (e.g., balance validation dashboards, recon gap tables), but the onboarding kick-off and step transitions must be text-only.
new customer, new account). If the request is ambiguous, ask: "Are you looking to run the full onboarding flow for a new entity, or do you just need help with a specific step (like uploading wallets or validating balances)?"
in the chat (not as an HTML widget or visualization) so they know what to expect. Use something like:
> Here's the onboarding roadmap for Entity Name]: > > 1. Upload Wallets — Add on-chain wallets and/or exchange accounts > 2. Data Collection (Commit) — Pull on-chain data for the uploaded wallets > 3. Validate Balances — Cross-check TRES balances against on-chain sources > 4. Reconciliation — Review and resolve any balance gaps > 5. Cost Basis — Configure and run cost basis calculation > 6. Export Unidentified Addresses — Extract 3rd-party addresses from transactions > 7. Import Contacts — Label and import the identified addresses back into TRES > 8. Rollup Rules — Set up transaction aggregation for high-volume wallets > > We'll go through each step together. Ready to start?
Work through each step in order. After completing each step, briefly summarize what was accomplished and ask the user if they're ready to move on. If a step isn't relevant (e.g., no high-volume wallets that need rollup rules), the user can skip it.
Important: For each step, invoke the corresponding skill by name. Read its SKILL.md and follow its instructions. The sub-skill handles all the details — your job is to transition smoothly between steps and keep the user oriented.
Skill: tres-wallets-upload
Invoke the tres-wallets-upload skill. This guides the user through adding their on-chain wallets and/or exchange accounts into TRES. Once wallets are created, confirm the count and proceed.
Transition: "Wallets are uploaded. Next, we'll collect on-chain data for them."
Skill: tres-data-collection-commit
Invoke the tres-data-collection-commit skill. This triggers on-chain data collection for the wallets that were just uploaded. The commit may take some time depending on the number of wallets and their history.
Transition: "Data collection is in progress / complete. Next, let's validate that the balances TRES pulled match what's actually on-chain."
Skill: tres-asset-balance-validation
Invoke the tres-asset-balance-validation skill. This compares TRES balances against DeBank (for EVM wallets) and generates a discrepancy report. Review any issues with the user.
Transition: "Balance validation is done. Let's move on to reconciliation to close any remaining gaps."
Skill: tres-recon-gaps
Invoke the tres-recon-gaps skill. This fetches reconciliation gap data, displays it, and helps the user resolve any on-chain vs. book balance discrepancies. If there are no gaps, celebrate and move on.
Transition: "Reconciliation is squared away. Now let's set up cost basis calculation."
Skill: tres-cost-basis
Invoke the tres-cost-basis skill. This walks through strategy selection if needed, triggers cost basis calculation, surfaces per-asset results and financial issues, and covers reevaluations, spec-ID rules, and related cost basis operations as needed. Review any financial issues (negative balances, missing fiat prices) that surface.
Transition: "Cost basis is configured and running. Next, let's identify the external addresses your wallets have been interacting with."
Skill: tres-export-3rd-party-contacts
Invoke the tres-export-3rd-party-contacts skill. This extracts all unidentified external addresses from the entity's transactions and exports them as an XLSX workbook. The user will need to review and label these addresses (who each address belongs to).
Transition: "The unidentified addresses have been exported. Once you've labeled them, we'll import them back as contacts. If you need time to fill in the names, we can pause here and come back to the next step later — just say 'continue onboarding' when ready."
Note: This step produces output that the user needs to review and fill in offline. Offer to pause the onboarding here and resume later. If the user wants to continue immediately (maybe they'll label contacts later), that's fine — skip to Step 8.
Skill: tres-import-contacts
Invoke the tres-import-contacts skill. This reads the user's labeled contacts file and imports the address-to-name mappings into the TRES address book.
Transition: "Contacts are imported! Last step — let's set up rollup rules for any high-volume wallets."
Skill: tres-rollup-rules
Invoke the tres-rollup-rules skill. This helps the user identify wallets with high transaction volume and create daily or monthly rollup rules to consolidate small sub-transactions (gas fees, staking rewards, etc.) into cleaner aggregated entries.
If no wallets need rollup rules, skip this step.
Once all steps are done (or skipped where appropriate), wrap up with a summary:
> Onboarding complete for Entity Name]! > > Here's what we accomplished: > - Uploaded X wallets across Y networks > - Collected on-chain data > - Validated balances (note any outstanding issues) > - Reconciled gaps (note status) > - Configured cost basis strategy: strategy] > - Exported and imported Z contacts > - Set up N rollup rules > > The entity is ready to go in TRES Finance.
The user may need to pause and come back later (especially around Step 6-7 where they need to label contacts offline). Keep track of which step was last completed. If the user comes back and says something like "continue onboarding" or "where were we with onboarding entity]?", pick up from the next incomplete step.
The user can skip any step if it's not relevant. For example:
Always respect the user's choice and move to the next step.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 16,499 | 10,219 | -38% | 1 | 1 | 0% | 1,695 | 2,719 | +60% | 0 | 0 | — |
case-02 | fail→pass | 11,861 | 10,166 | -14% | 1 | 1 | 0% | 1,007 | 2,608 | +159% | 0 | 0 | — |
case-03 | fail→pass | 8,593 | 5,382 | -37% | 1 | 1 | 0% | 1,067 | 2,655 | +149% | 0 | 0 | — |
case-04 | pass→pass | 9,614 | 10,092 | +5% | 1 | 1 | 0% | 1,501 | 2,635 | +76% | 0 | 0 | — |
case-05 | fail→pass | 55,785 | 14,311 | -74% | 1 | 1 | 0% | 3,495 | 3,058 | -13% | 0 | 0 | — |
case-06 | pass→pass | 27,930 | 27,221 | -3% | 1 | 1 | 0% | 4,041 | 5,004 | +24% | 0 | 0 | — |
case-07 | fail→pass | 7,287 | 8,915 | +22% | 1 | 1 | 0% | 935 | 2,301 | +146% | 0 | 0 | — |
case-08 | fail→pass | 36,858 | 11,761 | -68% | 1 | 1 | 0% | 6,709 | 2,831 | -58% | 0 | 0 | — |
case-09 | fail→fail | 8,107 | 3,092 | -62% | 1 | 1 | 0% | 1,221 | 2,315 | +90% | 0 | 0 | — |
case-10 | fail→fail | 13,645 | 4,231 | -69% | 1 | 1 | 0% | 1,987 | 2,350 | +18% | 0 | 0 | — |
case-11 | fail→fail | 12,396 | 3,603 | -71% | 1 | 1 | 0% | 1,779 | 2,410 | +35% | 0 | 0 | — |
case-12 | fail→fail | 8,953 | 9,647 | +8% | 1 | 1 | 0% | 1,221 | 2,426 | +99% | 0 | 0 | — |
case-13 | fail→fail | 7,802 | 4,167 | -47% | 1 | 1 | 0% | 1,124 | 2,470 | +120% | 0 | 0 | — |
case-14 | pass→pass | 11,911 | 4,915 | -59% | 1 | 1 | 0% | 1,661 | 2,748 | +65% | 0 | 0 | — |
case-15 | fail→fail | 10,434 | 5,191 | -50% | 1 | 1 | 0% | 1,492 | 2,635 | +77% | 0 | 0 | — |
case-16 | fail→pass | 10,259 | 6,299 | -39% | 1 | 1 | 0% | 1,373 | 2,667 | +94% | 0 | 0 | — |
case-17 | fail→pass | 9,042 | 3,406 | -62% | 1 | 1 | 0% | 1,223 | 2,445 | +100% | 0 | 0 | — |
case-18 | fail→pass | 9,623 | 4,775 | -50% | 1 | 1 | 0% | 1,319 | 2,504 | +90% | 0 | 0 | — |
case-19 | fail→pass | 14,062 | 6,720 | -52% | 1 | 1 | 0% | 1,904 | 2,901 | +52% | 0 | 0 | — |
case-20 | fail→pass | 14,219 | 5,942 | -58% | 1 | 1 | 0% | 1,986 | 2,707 | +36% | 0 | 0 | — |
case-21 | fail→pass | 13,246 | 4,112 | -69% | 1 | 1 | 0% | 1,768 | 2,458 | +39% | 0 | 0 | — |
case-22 | fail→fail | 9,665 | 9,387 | -3% | 1 | 1 | 0% | 1,199 | 2,916 | +143% | 0 | 0 | — |
case-23 | fail→pass | 18,806 | 3,791 | -80% | 1 | 1 | 0% | 2,533 | 2,359 | -7% | 0 | 0 | — |
case-24 | fail→pass | 9,531 | 3,258 | -66% | 1 | 1 | 0% | 1,070 | 2,333 | +118% | 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. 24 cases were attempted. The headline lift of +58 percentage points is the difference between those two pass rates over the 24 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.