Skip to content
Back to search

common-sense

Agent skill for practical judgment: time and place awareness, reading the user, intent, right-sized effort and clear reporting.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/sdtechsystems--common-sense/badge.svg)](/listings/sdtechsystems--common-sense)

HTML

<a href="/listings/sdtechsystems--common-sense"><img src="/api/listings/sdtechsystems--common-sense/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 81 lines
  1. <p align="center">
  2. <img src=".github/assets/banner.png" alt="SD Tech Systems" width="640">
  3. </p>
  4. # common-sense
  5. A skill that gives your AI agent the judgment of a good employee learning to work with a new boss: it notices the time, your working hours and which computer you are on, learns how you like to work and how much you know about code, asks when it should and acts when it can, and suggests the right model or effort level for the job instead of using the biggest one for everything.
  6. ```bash
  7. npx skills add sdtechsystems/common-sense
  8. ```
  9. Works with Claude Code and any agent that reads `SKILL.md` files.
  10. ## What changes
  11. **Without it**
  12. > **You (23:40):** migrate the users table to the new schema
  13. >
  14. > **Agent:** *runs the migration on the connected database*
  15. **With it**
  16. > **You (23:40):** migrate the users table to the new schema
  17. >
  18. > **Agent:** It's late and this touches production data, so I'll write the migration and a rollback, run it against the local copy, and leave the production run for you to approve. Done in about 10 minutes.
  19. A few more:
  20. - **Knows your machines.** On a laptop it hasn't seen, it asks if you'll use it regularly, offers to install what's missing, and reminds you what lives on your main PC, or that you're still signed in there.
  21. - **Adapts to you.** If you don't code, it guides you step by step, explains jargon once and stops when you've got it, and never hands you coding homework. If you do, it gets out of your way.
  22. - **Local common sense.** Examples, formats, holidays and tact that fit where you live.
  23. - **Asks the right amount.** Several questions when a request is unclear, one with options when it's partly clear, none when it already knows how you work.
  24. - **Asks for everything up front:** web access, installs, sign-ins and admin rights, so a long task doesn't stop halfway.
  25. - **Right-sizes the work:** suggests `/model haiku` for a one-line fix, or `/effort ultracode` for a 300-file migration, only when there's a clear mismatch.
  26. - **Checks official docs** for commands and APIs that change often, with your permission.
  27. - **Tells you what it built**, in a short brief in the chat, when the change isn't obvious.
  28. ## What's inside
  29. ```
  30. skills/common-sense/
  31. ├── SKILL.md first-use intro and the quick check before every task
  32. └── references/
  33. ├── context.md time, working hours, place, environment
  34. ├── devices.md main vs. secondary computers, device inventory
  35. ├── reading-the-user.md how you work and how much you know about code
  36. ├── local-sense.md local culture, examples and tact
  37. ├── intent.md when to act, assume, or ask one or several questions
  38. ├── anticipation.md asking for permissions and admin rights up front
  39. ├── resource-sense.md model, effort and cost (Claude Code and other agents)
  40. ├── staying-current.md checking official sources
  41. └── communication.md before / during / after, and the brief
  42. ```
  43. What it learns about you is kept in your agent's own memory, never in the repository.
  44. The agent only loads a reference file when it needs it, so the skill stays light on context.
  45. ## Limits
  46. - It suggests settings; it does not change them. You stay in control of models, permissions and spending.
  47. - It never looks up your location online. It only uses what your system already shows (time zone, language).
  48. - Command names in `resource-sense.md` were checked against the Claude Code docs on 2026-11-21. They may change; the skill tells the agent to verify when it can.
  49. ## Pairs well with
  50. [personality-traits](https://github.com/sdtechsystems/personality-traits): choose how your agent works (lean, diligent, cautious, thrifty, skeptic, explainer...).
  51. ## Security
  52. Markdown only: no scripts, no network calls, no access to secrets. See [SECURITY.md](SECURITY.md).
  53. ## License
  54. [MIT](LICENSE)
  55. ---
  56. Developed by **SD Tech Systems** · Part of SD Tech

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.