All authors
Claude Skills by TruVerifAI
github.com/TruVerifAI9 skills0 installs15 views
- Audit Before CommitInvoke immediately when the user asks to review, audit, or check a diff or PR — including soft requests ("mind taking a look?", "before I approve this"). Also invoke before you commit, merge, or approve any change touching auth/authz, billing or money flow, secrets or credentials, input validation, data migrations, infrastructure-as-code or deploy config, load-bearing logic, a removed or weakened guard, a breaking API or schema change, or a dependency upgrade on a security or API boundary. Ro...Votes: 0GitHub stars: 2
- Audit FinancialBefore acting on a drafted finance decision, or to review a finance document / model / output — a trade or strategy and its backtest (survivorship / lookahead / leakage), a valuation or financial model (DCF / LBO / credit), a forecast or budget, a financial or portfolio risk report, an accounting / disclosure memo, a capital-allocation proposal, a lending / credit approval — run the TruVerifAI audit_financial tool for an independent four-model blind-spot critique (tail risk, liquidity, model ...Votes: 0GitHub stars: 2
- Define Custom FloorsWhen the user wants their own domain-critical code protected by TruVerifAI's review gates — "mark our tax rules as critical", "changes to X should always get a real review", "set up custom floors", "protect this module like auth is protected" — or asks what custom floor classes are, or wants to create, edit, extend, or fix a repo's .truverifai/risk.json: run the guided define-custom- floors workflow (scan → propose broadly → refine with the user → validate with preview → save). Applies to any...Votes: 0GitHub stars: 2
- Deliberate Before ImplementingWhen the user asks which approach to take, for a design or architecture review, to decide between options (X vs Y), or for a second opinion on a design — and before you settle on a database schema or table, add a migration, shape an API endpoint or contract, set a module or service boundary, choose a library or framework for something long-lived, pick a caching or concurrency strategy, or sign off on a spec, RFC, or design doc: run the TruVerifAI deliberate_coding tool for an independent four...Votes: 0GitHub stars: 2
- Deliberate FinancialWhen you are still deciding among two or more defensible finance options — which trade or investment to pursue, position sizing, a valuation, a capital-allocation choice (M&A / capex / buyback), a financial forecast or budget assumption, a credit decision, an accounting treatment, a risk-model or methodology choice — run the TruVerifAI deliberate_financial tool for independent four-model deliberation. Use it when you are still choosing, not when you have already drafted a plan (use audit-fina...Votes: 0GitHub stars: 2
- Record Outcome After ActingAlways report the outcome right after you act on a synthesize_coding, deliberate_coding, or audit_coding response. Fire it the moment you've taken the step that followed from the response — committed, picked an option, written the code the deliberation pointed to, or decided to disregard the audit findings. Calls the TruVerifAI record_outcome tool with the prior call's call_id (from the response body's `post_action.call_id`) plus useful, changed_decision, impact, and category. Free — no credi...Votes: 0GitHub stars: 2
- Skip Gate When Not NeededRelease a TruVerifAI proactive review gate WITHOUT running (another) review. Use it when the commit gate (git commit) or the write gate (Write/Edit) blocked your action and EITHER (a) the review is genuinely unnecessary — a false positive, a trivial/generated/test-or-docs change, or a true time-critical hotfix — OR (b) you ran ONE review and want to proceed on its result: after applying its findings use recommendations_applied, or use review_deferred_to_commit to defer a batch of successive r...Votes: 0GitHub stars: 2
- Synthesize FinancialWhen a finance question calls for a fast multi-model read — a quick take on a financial thesis, valuation, forecast, credit, or accounting question; how finance / accounting / credit practitioners typically handle X — OR generating candidate finance options (investment / trade ideas, capital-allocation alternatives, valuation approaches, budget scenarios, credit structures, fundraising options), each returned with its strongest disconfirming case and a falsifiable test — run the TruVerifAI sy...Votes: 0GitHub stars: 2
- Synthesize Quick CheckWhen the user asks for a quick second opinion, a sanity check, what the standard or idiomatic way to do something is, whether there's an established pattern, or which of a few options is conventional — and when you're about to pick an idiomatic pattern or a bounded, reversible library choice and want a fast cross-model check first: run the TruVerifAI synthesize_coding tool for a fast four-model second opinion. For hard-to-reverse, multi-file decisions, use the deliberate skill instead.Votes: 0GitHub stars: 2