q-kdb-code-review
por beee003
Coding Agents & IDEsResumo
AI-powered code review for Q/kdb+ — catch bugs in the most terse language in finance.
Descrição
O que esta skill faz
Reviews Q/kdb+ code for type errors, performance antipatterns, and security vulnerabilities. Routes requests through Astrai's inference router, which selects models based on query complexity. Supports three strictness modes: standard, strict, and security.
General-purpose AI models treat Q's terse syntax as noise and miss finance-domain patterns like attribute gaps, adverb misuse, and asof join semantics, which this skill addresses through Q-specific prompting and kdb+ internals knowledge.
Quando usá-la
- Catching implicit type coercion bugs before production deployment
- Finding missing sorted attributes on asof join time columns
- Auditing .z.pg and .z.pw handlers for injection vulnerabilities
- Replacing each loops with vector operations in time-series calculations
- Detecting race conditions in .z.ts timer callbacks
Exemplo de fluxo de trabalho
User asks: Review tick.q in strict mode
- 1
Read tick.q from disk
- 2
Construct review prompt with strict mode parameters and Q semantics context
- 3
Route request to Astrai inference router for optimal model selection
- 4
Parse model response and categorize findings by severity
- 5
Format output with line numbers, explanations, and suggested fixes
Issues labeled CRITICAL, WARNING, or INFO, each with the affected line, explanation, and a concrete fix, followed by a count summary
Requisitos
- ASTRAI_API_KEY from as-trai.com (free tier available, paid plans for higher volume)
- Optional: API keys from AI providers (Anthropic, OpenAI, Google, DeepSeek, Mistral, Groq, Cohere, Perplexity) for BYOK routing
Instalação
clawhub install beee003/q-kdb-code-reviewExecute o comando da CLI ou cole o prompt no Claude Code, Cursor, Codex ou no agente de código que você usa.