Analysis: --log-for human|machine vs --output text|json design review

[Copilot]

Objective analysis of PR #1508's --log-for human|machine flag design vs traditional --output text|json approach.

Recommendation

Keep --log-for human|machine - audience-centric naming serves LLM-first use case better than format-centric alternatives.

Key Findings

Why --log-for Works

• LLM discoverability: Self-identifies with "machine" when reading --help, no format reasoning required
• Future-proof: Abstraction permits internal format evolution (JSON Lines → Protocol Buffers) without API breakage
• Semantic clarity: Communicates audience intent, not implementation detail
• Consistent: Aligns with logFor config property in devproxyrc.json

Acknowledged Tradeoffs

• Deviates from CLI conventions (kubectl -o, Azure CLI --output)
• Format not immediately explicit (requires docs/testing to know it's JSON Lines)
• Binary choice limits extensibility vs multi-format --output json|yaml|xml

Why Tradeoffs Are Acceptable

• Primary use case is LLMs/agents per Add --log-for human|machine flag for LLM/agent-friendly output #1507
• Dev Proxy is specialized tooling, not general-purpose CLI
• Reporters (JsonReporter/PlainTextReporter) already handle file output with format-centric naming

Implementation Issues Identified

From PR review comments:

1. LogFor property: read-only in IProxyConfiguration, settable in ProxyConfiguration (inconsistent)
2. Dead code: LogFor property in ProxyConsoleFormatterOptions unused by ProxyConsoleFormatter

Full Analysis

See /tmp/pr-1508-analysis.md for detailed pros/cons comparison, comparative matrix, and alternative approaches.

Original prompt

Get information about PR 1508. Review it objectively: does the --log-for human|machine option make sense or is it just unnecessary complication over --output text|json? Weigh pros and cons for either approach. Consider the fact that this is primarily meant for LLMs but could also be used by humands

The user has attached the following file paths as relevant context:

• .github/copilot-instructions.md

Created from VS Code.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.