agentleFS
Sign inStart free
effectustasieffectustasi/agent-receipts/skills/no-guessing/SKILL.md

no-guessing

Skill1 starsChanged 41 days agoMITView on GitHub

MANDATORY before any edit that adds a call, argument, or flag for a library or tool, and whenever a task says a library "supports" something. Load it first, then confirm the feature exists in the installed version (repo, package source, --help); docs and task text describe other versions. Never edit a pinned or vendored dependency to add what is missing.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    The installed version: read the package source or type definitions in node_modules, site-packages, vendor, or equivalent.

    line 22
  2. Distinctive:

    Do not silently write it anyway.

    line 30

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

No guessing

Memory is a hint, not a source. Before you use a name you have not seen in this session, look it up. That includes names you were told: a task, an issue, or a doc page saying "call client.upload(..., resumable=True)" is a claim to check, not proof it exists in the installed version.

Applies to

  • Functions, methods, classes, and their parameter names, from any library.
  • CLI commands and flags.
  • Config keys, environment variables, feature flags.
  • API endpoints, request and response fields.
  • File paths and module imports inside the project.

How to confirm (cheapest first)

  1. The repo: grep for an existing use. If the project already calls it, copy that usage.
  2. The installed version: read the package source or type definitions in node_modules, site-packages, vendor, or equivalent. Versions differ; the installed one is the truth.
  3. The tool itself: <cmd> --help, man <cmd>, python -c "import x; help(x.y)".
  4. Official docs for the installed version.

One lookup is enough. This is a reflex, not a research project.

When you cannot confirm

Do not silently write it anyway. Either pick something you can confirm, or mark it:

Uses `client.batch_upsert(...)`. **Unconfirmed**: could not find it in the installed SDK (v2.3).
If it fails, check the SDK changelog for the batch method name.

When the docs and the installed version disagree

The installed version wins.

  • Never edit a dependency to add what the docs promised: anything in node_modules, site-packages, vendor/, third_party/, or a package pinned in requirements.txt, package.json, go.mod and the like, even when its source sits in the repo. It is not your code. The next install erases the change, and the project then breaks.
  • Solve it in the project's own code, or say the upgrade it needs:
`DataFrame.map` only exists from pandas 2.1; the installed version is 1.5.3.
Used `DataFrame.applymap` instead, which 1.5.3 has. After upgrading pandas, switch to `map`.

Red flags you are guessing

  • You are about to write a parameter name "that it's probably called".
  • The API "looks like" another library you know.
  • You are writing a flag from a different version of the tool.
  • An error says the name does not exist and your next move is to try a similar name. Stop and look it up instead.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free