agentleFS
Sign inStart free
tinyorbit-aitinyorbit-ai/skills/skills/forge-dx/SKILL.md

forge-dx

Skill1 starsChanged 3 months agoMITView on GitHub

Live developer-experience audit of a built CLI, API, SDK or library — runs the getting-started flow, times hello-world, checks errors, help and docs, and scores it with evidence. Run by forge-review for dev-facing phases. Use when asked to "test the DX", for a "DX audit" or to "try the onboarding".

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the build isn't developer-facing, say so and exit.

    line 17
  2. Distinctive:

    First call / first command: run it for real. Capture stdout/stderr.

    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

forge-dx

Tests the experienced developer journey by living it, not by reading the plan (that's forge-harden's DevEx angle). For libraries, APIs, CLIs, SDKs, tools.

When it runs

  • Auto: forge-review's runtime pass invokes this when the build is developer-facing (CLI/API/SDK/lib), scoped to what the phase changed.
  • Standalone: full onboarding audit on demand.

If the build isn't developer-facing, say so and exit.

Process

  1. Be a first-time developer. From a clean state, follow only what a real developer would have: the README / docs / --help. No insider knowledge. Read wiki/learnings.md first — past DX rules are enforced.
  2. Run the real flow and instrument it:
    • Time-to-hello-world (TTHW): from "I have nothing" to "it did the thing once". Record the actual minutes and every step taken.
    • Install/setup: does it work as documented, exactly? Note every gap between docs and reality.
    • First call / first command: run it for real. Capture stdout/stderr.
    • Error experience: deliberately do it wrong (missing arg, bad input, no auth). Screenshot/paste the actual error messages. Are they actionable or cryptic? An unhelpful error is a finding.
    • CLI help / API surface: is --help / the signature self-explanatory? Naming consistent? Surprises?
    • Docs: can the core task be completed from docs alone? Where did you get stuck or have to read source?
  3. Scorecard with evidence. Rate each (TTHW, setup, first-success, errors, help/docs, naming) with the concrete observation behind the score — real numbers, real quotes, real screenshots. No score without evidence.
  4. Fix and capture. Objective defects (wrong docs, broken setup step, cryptic error with an easy better message, missing --help) → fix on the phase branch and re-verify. Subjective calls → one AskUserQuestion batch in the Decision Brief shape (../forge/references/question-style.md): concrete framing, named stakes, recommendation with the why. Append the rule-to-remember to wiki/learnings.md; deferred polish → wiki/improvements.md. Tell the user what you changed and captured.

Result

End with the FORGE_RESULT line from ../forge/references/headless.md.

Rules

  • Evidence-first: TTHW is a measured number, errors are pasted verbatim.
  • Real clean-state run — no "it would probably work".
  • Fix objective DX defects automatically; surface only taste.
  • Phase branch only; never ship here.
  • Judge respect-for-developer-time, never adoption/market.

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