lab5.ca blogNotes on cloud infrastructure, AI agents for business workflows, and web development.https://lab5.ca/Smoke-testing an LLM agent with a Claude Code skillhttps://lab5.ca/blog/where-pytest-stops-and-claude-code-starts/https://lab5.ca/blog/where-pytest-stops-and-claude-code-starts/A Claude Code skill that smoke-tests an LLM agent end-to-end — real Gmail, real Drive, real model. Verifies that the agent does the right thing on real inputs, which ruff and pytest cannot.Tue, 19 May 2026 00:00:00 GMTtestingclaude-codeai-codingsmoke-testMeasuring math-glyph token compressionhttps://lab5.ca/blog/measuring-glyph-compression/https://lab5.ca/blog/measuring-glyph-compression/I ran 30 SPEC.md rows through Claude's tokenizer to measure how much math-glyph notation compresses token count. Two numbers — ~30% encoding, ~90% reviewer-facing.Mon, 18 May 2026 00:00:00 GMTspec-driven-developmentclaude-codeai-codingbenchmarksCompressed spec-driven developmenthttps://lab5.ca/blog/spec-driven-development/https://lab5.ca/blog/spec-driven-development/I built pilot-spec to keep AI coding agents on one thread, with one spec, so I can actually track what changed, what was tested, and what broke.Wed, 13 May 2026 00:00:00 GMTspec-driven-developmentclaude-codeai-coding