Skip to content

oliverweissl/context-garden / skills/weeder

weeder

Verified

Measure and reduce a Skill's token footprint (always-loaded description + on-trigger SKILL.md body) without regressing routing accuracy or losing instructions. Use when a SKILL.md feels oversized, repeats rules, or has a vague description that might mis-route.

tests/fixtures/filler_sentences.json

View on GitHub
{
  "_comment": "Labelled sentences for we_filler.py. The smoke test requires 0 'instruction' sentences flagged and >= 80% recall on 'filler'. Instructions include tricky ones that contain filler-ish phrases (e.g. 'Note that you must...').",
  "filler": [
    "This is useful because it keeps the output readable.",
    "Feel free to experiment with the different subcommands.",
    "Note that it's important to understand the overall design.",
    "Hope this helps with your workflow.",
    "This skill is designed to make repository maintenance easier.",
    "It is worth noting that the report format changed over time.",
    "Please note that the tool was originally written for internal projects.",
    "Happy coding!",
    "Thanks for reading this guide.",
    "Basically, the script just wraps a few git commands.",
    "Keep in mind that this is a heuristic.",
    "It can be helpful to understand how the scoring works.",
    "This way, nothing important gets lost.",
    "The reason for this is that early versions were slow.",
    "You might want to look at the examples first.",
    "It's very helpful for large repositories.",
    "Good luck with your refactor!",
    "As an AI assistant, I can explain the internals in detail.",
    "Of course, results may vary between projects.",
    "This helps keep the context window small."
  ],
  "instruction": [
    "Never delete a data file without explicit user confirmation first.",
    "Run the smoke test before committing.",
    "Check whether the file is tracked by version control.",
    "Note that you must pass --out when optimizing in place.",
    "Always log every deletion for audit purposes.",
    "Do not edit files under references/ by hand.",
    "Don't skip the confirmation step.",
    "Only move sections classified as background or examples.",
    "You should keep the description under 60 words.",
    "Confirmation is required before any destructive operation.",
    "Ensure the routing accuracy does not drop.",
    "Use the --json flag for machine-readable output.",
    "Call test-function after every edit.",
    "Set the assist level in .context-garden/config.yaml.",
    "Avoid generic phrasing in the description.",
    "Prefer concrete trigger nouns over vague verbs.",
    "Please delete the temporary directory afterwards.",
    "Write one example prompt per real trigger.",
    "Keep the original skill until the diff report passes.",
    "Feel free to skip step 3 only when audit reports nothing to fix.",
    "Read the whole audit report before editing.",
    "Then compare the before and after accuracy.",
    "Restore any constraint that test-function reports as missing."
  ]
}
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall