agentsclimarketplace

Momentic explore prompt

Skill momentic-ai/skills/skills/momentic-explore-prompt

Generate an explore-prompt.md file that gives Momentic's explore agent (`momentic ai explore diff` / `momentic ai explore latest`) repo-specific context — which applications to test, the URLs tests must target, how to authenticate, where to save generated tests, and repo quirks. Use when setting up or improving the prompt file passed via `--prompt-file`.From its SKILL.md

Install
npx -y skills add momentic-ai/skills --skill momentic-explore-prompt

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 13 stars13 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.7 KB, ~1.3k tokens by cl100k_base, as published. Nobody here has run it

Generating an explore-prompt.md

What this file is

The explore agent (momentic ai explore diff over a git range, or momentic ai explore latest over the whole app) identifies changed user journeys and builds Momentic tests for them. Both accept extra instructions appended to the explorer's system prompt via two repeatable options:

  • --prompt <text> — inline text. Repeatable.
  • --prompt-file <path> — a file whose contents are appended. Repeatable.

Passing any --prompt or --prompt-file replaces the project's cloud custom prompt. The combined instructions are all --prompt-file contents followed by all --prompt values, each in the order passed — so multiple flags stack rather than one overriding the other.

explore-prompt.md is the conventional file passed to --prompt-file. It is repo-specific: it tells the explorer which apps exist in this repo, the URLs its tests must target, how to sign in, exactly where to save generated tests, and any quirks it must respect. It is NOT a place to restate general Momentic behavior — that already lives in the momentic-test skill.

Usage:

momentic ai explore diff --prompt-file ./explore-prompt.md

Gather the repo facts first

Do not write the prompt from memory. Inspect the repo and confirm each fact:

  1. Applications under test. Enumerate every frontend/app the generated tests will drive. For each, determine the URL the tests must hit. This is the served/deployed URL used at run time, which is often NOT the Vite/dev URL used while authoring locally. Cross-check momentic.config.yaml environments[].baseUrl and how each app is started in CI (the workflow that runs momentic ai explore diff).
  2. Authentication. How should the explorer sign in? Prefer a reusable login module — record its exact name and id. Note access assumptions (e.g. "this user can reach all parts of the app").
  3. Test placement. The folder each app's tests belong in, expressed relative to the workspace, including any mandatory path prefix. Make the constraints unambiguous.
  4. Throwaway / nested tests. Where tests created as a side effect of an explore run must be saved (usually the single gitignored momentic/junk/ folder). Confirm which path is gitignored.
  5. Quirks. Externally-persisted data that survives runs (must be randomized per run to avoid collisions), auth peculiarities, flaky areas, base-URL caching flags, etc.

Structure to produce

Write concise, imperative markdown. Recommended sections:

  • Applications and target URLs — one bullet per app: name — the URL tests must target — purpose — how it is served/run — the login module (name + id) if auth is required.
  • Journey → application mapping — how to map an in-scope diff to the app(s) it affects so the right editor/port is driven.
  • Test placement rules — the destination folder per app, with explicit HARD RULE callouts for any mandatory prefix. The explorer will otherwise save tests in the wrong place.
  • Throwaway / nested test rule — the gitignored junk folder that every nested test must land in, and that naming a file junk-* does NOT count.
  • Quirks to watch out for — the repo-specific gotchas gathered above.

Rules for a good prompt

  • Target URLs must be the URLs tests actually run against, never the authoring dev-server URLs.
  • Be explicit about path constraints and use HARD RULE: callouts — path mistakes are the most common and most costly explorer failure.
  • Keep it repo-specific and short. Omit anything already covered by general Momentic docs/skills.
  • Reference login modules by both name and id so the explorer can pin the exact module.

Reference template

Below is a placeholder-only skeleton with the sections a good prompt has. Fill in the bracketed values with your repo's real apps, URLs, modules, and folders — do not ship the placeholders.

This repo has <N> frontend application(s). When tests run, each app is served at
the URL below — target these, not the dev-server URLs used while authoring.

- <app-name> — <one-line purpose> at <https://url-tests-must-target>. Sign in
  first using the module named `<login-module-name>` (id: <login-module-id>).
  <access assumptions, e.g. this user can reach all parts of the app>.
- <app-name-2> — <purpose> at <https://url-2>. Tests MUST use <https://url-2>
  as their base URL.

When identifying changed user journeys, map each in-scope diff to the
application(s) above so the right app/port is driven.

Place new tests in the folder matching the app they drive. Every target path is
relative to your workspace and the `<mandatory-prefix>/` segment is MANDATORY —
never drop it:

- <app-name> journeys → <mandatory-prefix>/<app-folder>/.
- <app-name-2> journeys → <mandatory-prefix>/<app-2-folder>/.

HARD RULE — keep the `<mandatory-prefix>/` prefix on the authored test. Saving
to any path that omits it is a hard failure.

HARD RULE: any test created as a side effect of building another test is
throwaway and MUST be saved into the gitignored `<junk-folder>/` directory.
Naming a file `junk-*` does NOT count — only that directory is gitignored.

Some quirks to watch out for:

- <quirk, e.g. externally-synced data persists across runs — randomize any data
  you write each run so runs don't conflict>.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most test skills give in ~1.3k tokens

Counted across 1,201 of the 2,096 authors here whose files we hold, read 2026-09-06

  • Write a failing test before writing codein 43 of 1201, across 36 files
  • Run the full test suitein 36 of 1201, across 35 files
  • Test only one variable per experimentin 34 of 1201, across 17 files
  • Read product marketing context before asking questionsin 34 of 1201, across 14 files
  • Mock external dependenciesin 34 of 1201, across 30 files
  • Define primary, secondary, and guardrail metricsin 33 of 1201, across 16 files
  • Pre-determine sample size before startingin 31 of 1201, across 14 files
  • Test behavior rather than implementationin 31 of 1201, across 29 files
  • Formulate a hypothesis before designing a testin 30 of 1201, across 13 files
  • Document every test hypothesis, variant, and resultin 29 of 1201, across 11 files
  • Use descriptive test function namesin 25 of 1201, across 21 files
  • Commit to the methodology without stopping earlyin 24 of 1201, across 8 files

Said here and by no other author read

  • Inspect the repository to confirm all facts
  • Enumerate every frontend application under test
  • Determine the runtime URL for each application
  • Identify the reusable login module name and id
  • Define the relative test placement folder for each app
  • Specify the gitignored folder for throwaway tests

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.