AGENT SKILL

repl-code-execution

暂无简介

License 许可
Apache-2.0
最近更新
2026-11-14
Stars
★ 13
Forks
2

03 SKILL.md 正文

/repl-code-execution

When to Use

  • Build interactive REPL sessions for testing snippets without full compilation cycle
  • Incremental evaluation: execute lines and preserve variable state across calls
  • Agent-driven code probing: try expressions, inspect results, backtrack on error
  • Jupyter-style cell execution for compiled-language notebooks

Do NOT use for

  • Production code execution (use subprocess with strict sandboxing [[runtime-sandbox-runc]])
  • JavaScript eval (use vm.Script with [[runtime-metrics-meters]] for memory limits)

Node.js REPL with persistent context

import repl from 'repl'
import vm   from 'vm'

// Start a REPL server with shared context
const replServer = repl.start({
  prompt:  '> ',
  useColors: true,
  terminal: false,           // disable TTY features for programmatic use
  eval: customEval,
})

// Inject shared state into REPL context
replServer.context.db     = dbConnection
replServer.context.agents = agentRegistry

function customEval(cmd, context, filename, callback) {
  try {
    const result = vm.runInContext(cmd, context, {
      filename,
      timeout: 3000,           // 3s hard limit per expression
      breakOnSigint: true,
    })
    callback(null, result)
  } catch (err) {
    // Recover from incomplete input (multi-line expressions)
    if (isIncomplete(err)) {
      callback(new repl.Recoverable(err))
    } else {
      callback(null, `[error] ${err.message}`)
    }
  }
}

function isIncomplete(err: Error): boolean {
  return err.message.includes('Unexpected end of input') ||
         err.message.includes('Unexpected token')
}

Sandboxed expression evaluation (one-shot)

import vm from 'vm'

function evalExpression(code: string, context: Record<string, unknown> = {}): unknown {
  const sandbox = vm.createContext({
    console: { log: (...a) => results.push(a) },
    ...context,
  })
  const results: unknown[][] = []

  const script = new vm.Script(code, { filename: 'repl-eval' })
  const value  = script.runInContext(sandbox, { timeout: 2000 })

  return { value, logs: results }
}

// Usage:
const { value, logs } = evalExpression('1 + 2 * Math.PI', { Math })
// value = 7.283...

Incremental state REPL (evcxr-style session)

// Stateful session: each submission builds on previous bindings
class ReplSession {
  private ctx = vm.createContext({ require, console, process })
  private history: string[] = []

  eval(code: string): { result: unknown; error?: string } {
    this.history.push(code)
    try {
      const script = new vm.Script(code)
      const result = script.runInContext(this.ctx, { timeout: 5000 })
      return { result }
    } catch (err: any) {
      return { result: undefined, error: err.message }
    }
  }

  reset() {
    this.ctx = vm.createContext({ require, console, process })
    this.history = []
  }
}

const session = new ReplSession()
session.eval('const x = 42')
session.eval('const y = x * 2')      // x is available from previous eval
const { result } = session.eval('y') // → 84

Anti-Fake-Pass Checklist

❌ vm.runInContext without timeout → infinite loop hangs the process
❌ vm.createContext with full global → sandbox escape via process.exit or require
❌ No error recovery → multi-line input discarded on partial parse
❌ Shared context between untrusted sessions → state pollution across users
❌ eval() instead of vm.Script → no timeout, no sandbox boundary
❌ REPL context mutation without clone → downstream sessions see dirty state

04 同仓库更多 Skills

Horonomos-laps
Horonomos-laps/Yana-AI

openai--sentry--sentry

Use when the user asks to inspect Sentry issues or events, summarize recent production errors, or pull basic Sentry health data via the Sentry API; perform read-only queries with the bundled script and require `SENTRY_AUTH_TOKEN`.

★ 13
Horonomos-laps
Horonomos-laps/Yana-AI

ito-basket-compare

Compare Itô prediction-market baskets against a user's knowledge base, portfolio notes, financial context, watchlist, or research thesis. Use for read-only basket comparison and gap analysis without investment advice or live trading.

★ 13
Horonomos-laps
Horonomos-laps/Yana-AI

implementing-network-traffic-baselining

Build network traffic baselines from NetFlow/IPFIX data using Python pandas for statistical analysis, z-score anomaly detection, and hourly/daily traffic pattern profiling

★ 13
Horonomos-laps
Horonomos-laps/Yana-AI

detecting-evasion-techniques-in-endpoint-logs

Detects defense evasion techniques used by adversaries in endpoint logs including log tampering, timestomping, process injection, and security tool disabling. Use when investigating suspicious endpoint behavior, building detection rules for evasion tactics, or conducting threat hunting for stealthy adversary activity. Activates for requests involving evasion detection, defense evasion analysis, log tampering detection, or MITRE ATT&CK TA0005.

★ 13
Horonomos-laps
Horonomos-laps/Yana-AI

implementing-microsegmentation-with-guardicore

Implementing microsegmentation using Akamai Guardicore Segmentation to map application dependencies, create granular network policies, visualize east-west traffic flows, and enforce least-privilege communication between workloads across data centers and cloud.

★ 13
Horonomos-laps
Horonomos-laps/Yana-AI

ramdisk-tmpfs-patterns

暂无简介

★ 13