-
Backend skills: choosing API and database help without paying for clutter
We measured 12,025 backend skills. The median preloads 58 tokens, the heaviest 720, before your API task starts. Here is how to choose one.
-
Slash commands: choose the right shape for a named task
42,815 slash commands measured: compare the 17-token session cost, security scan, originals and copies, then choose the right command for named work.
-
Plugins: what one install brings into your session
We measured 16,285 Claude Code plugins. Learn what one installation brings, what the scan found, and why its per-session token cost may be unknown.
-
Settings permissions: know what you are opening
2,888 settings measured: see what permissions open, what the scan flags, and which project files are worth reading before you install one.
-
Claude Code skills: the preload price before you use one
Claude Code skills can cost 58 tokens before use. We measured 247,505 skills, scanned their contents, and picked six useful starting points.
-
Subagents: choosing focused workers outside your context
A practical guide to choosing subagents, the 43-token median, where their instructions live, what the scan found, and when another mod fits better.
-
Code review skills: what they cost and which ones read the diff
Code review skills measured: compare the 54-token median, full-session cost, scan grades, findings, originals and the best diff readers.
-
Copilot instructions: what your first prompt inherits
Copilot instructions: 2,056 files measured, with a 683-token median per session, scan findings, originals and practical starting points.
-
Data and AI skills: choose one for notebooks and models
7,868 data and AI skills measured: compare session cost, scan findings, copies and fit for notebook, model and data work before installing.
-
Debugging skills: choose a method without paying for it all day
Debugging skills can cost 55 tokens every session even when unused. Compare scan grades, risky lines, copies, and five ways to investigate bugs.
-
Design skills: choose the right help for your interface
11,027 design skills measured: session cost, scan grades, findings, copies, and six starting points for choosing interface help in real work.
-
DevOps skills for deployment, CI and cloud work
15,154 DevOps skills measured for deployment, CI and cloud work, with costs, scan findings and practical picks for your next release.
-
Frontend skills: choosing what loads for component work
We measured 11,514 frontend skills. See what loads every session, what the scan found, and which six are sensible starting points.
-
GEMINI.md: the file Gemini CLI pays for every session
GEMINI.md costs a median 512 tokens every Gemini CLI session. See what the scan found, which files are heaviest, and five worth studying first.
-
Git skills: commits, branches and pull requests worth the cost
12,055 Git skills measured: the median session load is 51 tokens, the 90th percentile 120, and six are worth a closer look.
-
Originals and copies: find the first skill
15,698 skills measured: learn to spot the original, read copy counts, compare token preload, and choose where a copied file deserves your trust.
-
Planning skills: choose a plan before code
26,049 planning skills measured: how to choose a plan before code, what each costs per session, and when a rule or hook fits better.
-
Security skills: who is checking the checker?
Security skills can inspect your code, but the checker also costs tokens and may ask for access. We measured 15,178 to show the trade-off.
-
Testing skills: choose one that writes tests or runs them
13,825 testing skills measured: choose a skill that writes or runs tests, see token cost and scan grades, then start with six grounded picks.
-
Skill, command, rule or hook: choose the right mod for the job
Skill, command, rule or hook? We measured 348,758 mods to show what loads, what runs, what it costs, and when each kind fits.
-
Documentation mods for changelogs and technical writing
17,683 writing mods measured: costs, scan grades, and practical picks for changelogs, references, and project documentation in coding repositories.
-
What an AGENTS.md file costs you every session
34,245 instructions files measured. What AGENTS.md, CLAUDE.md and copilot-instructions.md cost on every session, how large projects keep theirs, and what such a file cannot do.
-
Claude Code hooks: read the command before you install it
8,472 Claude Code hooks measured: which event each one fires on, what the scan found in the 27 graded C or D, and six worth reading before you install anything.
-
Cursor rules: the one setting that decides what they cost you
We measured 14,110 Cursor rules. Two in three are read into every chat whether you need them or not. Here is what that costs, what the scan found, and five rules worth a look.