Index of every page on this site: llms.txt
All skills
erkamyaman avatar

/overnight-optimization

@7fcc555
Run it yourself

Use when you want an AI agent to improve performance, size or startup time of a project over a long unattended run, with tests as a safety gate.

  • 2 files
  • 2.2 KB
  • Updated 2 months ago
  • GitHub
Use this Skill: https://skilld.dev/gh/erkamyaman/dhh-p-bloom-agent-skills/overnight-optimization

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈42 tokens for metadata: the name and description. ≈365 when used: this file. ≈138 more on demand in 1 file.

Description uses 2.1% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈169 characters, or 2.1%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Overnight optimization

Inspired by DHH's Rails World 2026 keynote: https://www.youtube.com/watch?v=vDjW_dRyKXY

From the talk

DHH says every optimization is now within reach: models can run overnight and deliver improvements of roughly 10x to 30x (per the auto-generated subtitles). He points to his half-megabyte Hype presentation app and the large CPU and memory savings estimated for Hey's Rust backend. The guardrails, log and steps below are our additions, not from the talk.

The talk's point: optimizations that no human had time for are now within reach if you let an agent work through them unattended. The risk is breaking behaviour, so the run needs a measurable target and a hard safety gate.

When to use

  • A project has a clear metric (bundle size, startup time, request latency, memory).
  • You have tests that give you real confidence.

Steps

  1. Measure the baseline and save the numbers to a file.
  2. Define the target ("cut bundle size by 20%") and the guardrails ("all tests pass, no visible behaviour change").
  3. Tell the agent to work in small steps: one change, run tests, run the benchmark, keep or revert.
  4. Have it log every attempt (what changed, before and after numbers, kept or reverted).
  5. Run it in its own branch.

Verification

In the morning, re-run the tests and the benchmark yourself and compare to the baseline. Read the log and review every kept change before merging.

See references/run-log.md.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at 7fcc555. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last month.

Activeupdated 2 months ago

README badge

README badge for erkamyaman/dhh-p-bloom-agent-skills/overnight-optimization