Home/Catalog/Product
Product — what could this be?
Brief 107

In-App Search Audit

The product's own search — whether users can find what they are looking for: relevance, recall, forgiveness, and what happens on zero results.

SEARCH.md4 phases7 lenses~2.6k chars

Free & open · no signup · read-only — it ends by asking · nothing leaves your machine

What it does

You are working inside this repo. Mission: judge the product's built-in search as a user experiences it — can they find what they came for, is the best result first, and does the product help when nothing matches. This is the in-product find feature, not SEO.

Read-only pass. Run real queries against the product, read the search and indexing code; change nothing but the report file.

What it looks for

7 lenses it looks through

Phase 2 sweeps the codebase through every one of these, citing file and line for each finding.

1

Relevance

do the best matches rank first, or is ordering arbitrary

2

Recall & coverage

is everything findable that should be; fields or content missing from the index

3

Forgiveness

typos, synonyms, partial matches, case and accent folding

4

Zero-result states

a helpful path forward versus a blank dead end

5

Speed & feedback

responsiveness, as-you-type feedback, a loading state

6

Filters & scoping

useful facets to refine a broad query

7

Query understanding

multi-word queries, phrases, and intent, not just substring match

The deliverable

What lands in your repo

One structured report at the repo root — or in reports/, if you keep one — the same shape every time, ready for a teammate — or the optional Studio — to act on.

SEARCH.md
1
Query runs
the test queries, what they returned, and the verdict on each
2
Findings
each: lens · query · what failed · the fix
3
Recall gaps
the content that should be findable and is not
4
Zero-result design
what the product should offer when nothing matches
One file. Evidence-backed. It ends by asking before touching anything.
How it works

What it makes the agent do

Every Goal Prompt follows the same four steps, so results are consistent and repeatable — no matter which one you run.

1
Phase 1

Search like a user

Run a set of realistic queries.

2
Phase 2

Audit through 7 lenses

Cite the query and result for every finding.

3
Phase 3

Curate

Rank by how often the query type is used and how badly it fails.

4
Phase 4

Report

Create SEARCH.md at repo root.

Get started

Three ways to run this Goal Prompt

01 · COPY

Paste it in

Copy it and paste it into your agent inside the repo you want checked.

02 · INSTALL

As a slash command

Install the goal plugin once — two commands — then just type /goal:in-app-search.

1/plugin marketplace add GhostlyGawd/goal-prompts
2/plugin install goal@goal-prompts

Or install only this Goal Prompt as /goal-in-app-search:

curl -fsSL https://goal-prompts.vercel.app/install | BRIEF=107 sh
03 · AGENT

From an agent (MCP)

Let an agent fetch it mid-conversation, or pull the raw Goal Prompt by URL.

https://goal-prompts.vercel.app/raw/107.md
Transparency

The exact prompt

Nothing hidden — this is the whole Goal Prompt, verbatim. Read it in a minute, edit it, or copy it as-is.

Read the full brief (2,560 characters)
# Goal: In-App Search Audit

You are working inside this repo. Mission: judge the product's built-in search as a user experiences it — can they find what they came for, is the best result first, and does the product help when nothing matches. This is the in-product find feature, not SEO.

Read-only pass. Run real queries against the product, read the search and indexing code; change nothing but the report file.

## Phase 1 — Search like a user
- Run a set of realistic queries: exact, partial, misspelled, multi-word, and ones that should return nothing.
- Note what comes back, in what order, and how fast.
- Find what is indexed and what is silently unsearchable.

## Phase 2 — Audit through 7 lenses
Cite the query and result for every finding.
1. **Relevance** — do the best matches rank first, or is ordering arbitrary
2. **Recall & coverage** — is everything findable that should be; fields or content missing from the index
3. **Forgiveness** — typos, synonyms, partial matches, case and accent folding
4. **Zero-result states** — a helpful path forward versus a blank dead end
5. **Speed & feedback** — responsiveness, as-you-type feedback, a loading state
6. **Filters & scoping** — useful facets to refine a broad query
7. **Query understanding** — multi-word queries, phrases, and intent, not just substring match

## Phase 3 — Curate
- Rank by how often the query type is used and how badly it fails.
- For each, name the fix — index a field, add synonyms, fix ranking, design the empty state.
- Separate "wrong results" from "no results handling"; both lose the user.

## Phase 4 — Report
Create `SEARCH.md` at repo root:
1. **Query runs** — the test queries, what they returned, and the verdict on each
2. **Findings** — each: lens · query · what failed · the fix
3. **Recall gaps** — the content that should be findable and is not
4. **Zero-result design** — what the product should offer when nothing matches

Start the report with today's date. If `SEARCH.md` already exists from a previous run, read it first and lead with what changed since.

## Rules
- Judge by real queries and their results, not by the search box's looks
- A confident wrong result is worse than an honest empty state
- No in-app search in this repo? Say so in a one-paragraph null report and stop — a null result is a valid finding.
- If a `reports/` directory exists at the repo root, write the report there instead of the root.
- Before asking, present the top findings as a ranked list in plain words
- Report only — end by asking which search fixes to make first

House rules for this Goal Prompt

  • Judge by real queries and their results, not by the search box's looks
  • A confident wrong result is worse than an honest empty state
  • No in-app search in this repo? Say so in a one-paragraph null report and stop — a null result is a valid finding.
  • If a reports/ directory exists at the repo root, write the report there instead of the root.
  • Before asking, present the top findings as a ranked list in plain words
  • Report only — end by asking which search fixes to make first
Keep exploring

More Product briefs