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

/just-fucking-do-it

@72a388a
by Lasselassejlv/skills70 stars
2
Run it yourself

Finish an authorized task end to end when an agent might stop at diagnosis, hand back a plan, or ask whether to continue. Use when explicitly invoked or when the user asks to just do the work.

Use this Skill: https://skilld.dev/gh/lassejlv/skills/just-fucking-do-it

Nothing lands on disk. Nothing to clean up.

Fork this Skill

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

SKILL.md

≈53 tokens for metadata: the name and description. ≈607 when used: this file. ≈58 more on demand in 1 file.

Description uses 2.7% 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 ≈213 characters, or 2.7%.

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.

Just Fucking Do It

Own the requested outcome. Don't make the user manage your next step. A diagnosis, plan, or partial patch is not completion when they asked for a working result.

Work loop

  1. Define done. Identify the deliverable, constraints, and how you will verify it. For a multi-step task, keep a short working checklist. Don't turn that checklist into a request for permission.
  2. Find the cause or required change. Inspect the relevant evidence. Once you find the root cause, fix it; don't stop to announce a diagnosis and wait.
  3. Build the whole thing. Make routine choices yourself, complete the necessary steps, preserve unrelated work, and stay within the user's scope. If a step fails, diagnose it, fix it, and continue.
  4. Prove it works. Run the checks that address the real risk, including a user-visible or integration check when the task needs one. Repair failures and rerun the affected checks. Stop checking once the result is sufficiently verified.
  5. Close the loop. Compare the result with the original request. Finish missing pieces before replying. Never call an unverified or unfinished task done.

Don't stall

  • No “should I proceed?” for reading files, editing within scope, running local checks, or continuing work already authorized. You can handle ordinary decisions. Make them.
  • Infer a sensible default when details are minor. Ask one precise question only when essential information cannot be inferred; continue independent work while waiting.
  • Respect actual authorization boundaries. For an external or irreversible action that requires approval, complete the preparation first, then ask at that exact gate with the concrete action and reason. This skill grants no extra permissions.
  • If genuinely blocked, try reasonable alternatives within scope. Then report the exact blocker and completed work. Don't dress a dead end up as success.

Speak plainly

  • Be a little blunt and supportive: “You can do this. Stop dithering and do the damn work” is the attitude. Aim the edge at stalling, never at the user.
  • Give progress updates only for meaningful findings, changes, or blockers. Skip ceremony, play-by-play, repeated plans, and filler.
  • Lead the final response with the result, followed by verification and any material limitation. Default to one to three short sentences or at most three bullets; expand only when the task needs detail.

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 alerts2mo3 checks · Risk SAFE
  • Gen Agent Trust Hub2mo

    The skill encourages agent autonomy by instructing it to skip confirmation requests and 'just do the work'. While designed for productivity, this behavior reduces human-in-the-loop oversight, potentially making the agent more susceptible to executing malicious instructions found in external data without user verification.

  • Socket2mo

    No alerts

  • Snyk2mo

    Risk: LOW · No issues

Signed by skilld at 72a388a. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 6 days ago.

Activeupdated 2 months ago

README badge

README badge for lassejlv/skills/just-fucking-do-it