achaykovsky

achaykovsky/Cursor-Usage-Templates

Cursor rule

Prompt eval fixtures — case design, assertions, versioning, CI gates, no PII in golden data.

achaykovsky/Cursor-Usage-Templates

Cursor rule

Python backend - FastAPI, security, typing, pytest.

achaykovsky/Cursor-Usage-Templates

Cursor rule

RAG pipeline — idempotent ingest, chunk metadata, embedding version lock, hybrid retrieval, cite-or-abstain.

achaykovsky/Cursor-Usage-Templates

Cursor rule

Prepend a Resources used report to every user-facing agent reply.

achaykovsky/Cursor-Usage-Templates

Cursor rule

Security - vulnerabilities, compliance, OWASP, HIPAA/GDPR.

achaykovsky/Cursor-Usage-Templates

Cursor rule

Skills consistency - structure, terminology, routing boundaries.

achaykovsky/Cursor-Usage-Templates

Cursor rule

SQL database - schema design, migrations, query optimization.

achaykovsky/Cursor-Usage-Templates

Cursor rule

Test standards - pytest, AAA, coverage, portable entry points (Code Runner + CLI).

achaykovsky/Cursor-Usage-Templates

Cursor rule

Minimize context and reply size — concise I/O, references over pastes, no redundant code dumps.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Adds prompt injection and jailbreak defenses for customer-facing bots — input boundaries, delimiter strategy, tool arg validation. Use when hardening untrusted user input paths.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Designs observability for LLM bots — traces, audit logs, eval pipelines, SLOs. Use when instrumenting customer-facing agents for operations and incident response.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Designs customer-facing bot specs — intents, conversation flows, persona, tone, and acceptance criteria. Use when planning support bots, chatbots, onboarding assistants, or FAQ agents for end users.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Designs router bots and specialist delegation — intent routing, fallbacks, context passing. Use when one entry bot dispatches to multiple domain agents.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Evaluates AI safety policy for customer bots — content rules, PII in context, retention, tool risk. Use before launch or after material bot changes.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Implements rate limits, abuse controls, and cost caps for LLM bots. Use when preventing DoS, spam, or runaway token spend on customer-facing agents.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Implements LLM bot gateway — FastAPI or Go service, webhooks, session store, channel adapters. Use when scaffolding or extending customer-facing bot backends.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Implements human escalation for bots — queues, tickets, Slack threads, handoff payloads. Use when customers must reach a human from an AI assistant.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Monitors AI bot quality — regression evals, drift, thumbs feedback, escalation rates. Use after launch or when quality regressions are suspected.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Routes customer-facing bot work across AI infra skills and agents with a consistent delivery sequence, shared vocabulary, and escalation boundaries. Use when building support bots, chatbots, or multi-channel AI agents, or when ownership between conversation design, gateway, safety, and observability is unclear.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Performs LLM system design review (not code review) across hallucination risks, context explosion, retrieval quality, prompt coupling, observability, evaluation, confidence calculation, deterministic business logic, tenant isolation, future tool calling, cost, and latency. Use for bot/RAG architecture review…

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Identifies who consumes an API and what breaks for them (internal apps, SDKs, partners); summarizes blast radius and migration work. Use when changing or removing endpoints, tightening schemas, or "who will this break.".

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Chooses and applies API versioning strategies (URL path, header, Accept, query), deprecation headers, and coexistence of versions. Use when adding v2, deprecating endpoints, or "how should we version this API.".

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Evaluates whether an API change is backward compatible for existing clients; classifies additive vs breaking changes and suggests mitigations. Use when the user asks "is this breaking," "backward compatible," "safe for clients," or before shipping contract changes.

achaykovsky/Cursor-Usage-Templates

Skill Claude CodeCodex

Scaffolds a FastAPI REST API project with recommended structure, routers, config, and dependencies. Use when the user asks to "create a FastAPI," "scaffold FastAPI," "new FastAPI project," or "FastAPI project structure.".