All jobs

Add Full-Text Search to an Axum API

Yes. An agent adds a Postgres tsvector column and GIN index by SQLx migration, a `q` filter on the Axum article list, a test, and opens a pull request.

Run this on RushOpens the console with this prompt ready to send, in a Project that has your repository.

What the agent does

The agent starts with the schema and the query it has to extend, because in a SQLx project the two are checked against each other at compile time. In an API shaped like launchbadge/realworld-axum-sqlx, Cargo.toml pins axum 0.3.4, sqlx 0.5 with the postgres, uuid and time features, and tower-http 0.2.0. migrations/1_setup.sql enables uuid-ossp, defines trigger_updated_at() and a case_insensitive collation; migrations/4_article.sql creates article with slug, title, description, body and a tag_list text[] column that already has a GIN index, article_tags_gin. The list endpoint lives in src/http/articles/listing.rs: ListArticlesQuery takes tag, author, favorited, limit and offset, and list_articles runs one sqlx::query_as! with a where clause of $n is null or … branches, ordered by article.created_at desc.

It adds migrations/5_article_search.sql, numbered after the existing four so sqlx migrate run applies it last. The migration adds a stored generated column, search tsvector generated always as (setweight(to_tsvector('english', title), 'A') || setweight(to_tsvector('english', description), 'B') || setweight(to_tsvector('english', body), 'C')) stored, and a GIN index on it. A generated column keeps the vector current on every insert and update without a trigger, which matters here because trigger_updated_at already fires on the table and the agent does not want two triggers racing on the same row.

In listing.rs it adds q: Option<String> to ListArticlesQuery and one more branch to the existing where: $7::text is null or search @@ websearch_to_tsquery('english', $7). websearch_to_tsquery accepts what a person types, quoted phrases and a leading minus included, and never raises a syntax error the way to_tsquery can. The ordering stays created_at desc when q is absent; with a term present the agent orders by ts_rank(search, websearch_to_tsquery('english', $7)) desc first, inside the same query so the query_as! macro still sees one statement. Because sqlx 0.5 has no offline metadata file in this repository, the agent starts Postgres in the sandbox, exports DATABASE_URL as .env.sample shows, runs sqlx db setup, and only then does cargo build succeed with the new column known to the macro.

The repository has no unit tests of its own; .github/workflows/rust.yml runs cargo build --verbose and cargo test --verbose against a Postgres service, then starts the server and runs the Realworld Postman collection with ./run-api-tests.sh. The agent adds tests/search.rs, an integration test that connects with DATABASE_URL, inserts two articles through the users and articles routes, calls GET /api/articles?q=… and asserts that only the matching slug comes back. It runs cargo fmt --check, cargo build, cargo test and a few curl calls against the running server in the sandbox, and pastes the responses in the pull request.

Start it with one API call

POST https://api.phoenix.so/o/{org}/sessions with this body:

{
  "agent": "claude",
  "input": "Add full-text search to the article list of this Axum API and open a pull request. Read Cargo.toml, migrations/1_setup.sql, migrations/4_article.sql, src/http/articles/listing.rs, src/http/articles/mod.rs, src/http/mod.rs, .env.sample and .github/workflows/rust.yml first. Add migrations/5_article_search.sql with a stored generated tsvector column named search over title (weight A), description (weight B) and body (weight C), and a GIN index on it. Add an optional q query parameter to ListArticlesQuery in listing.rs and extend the existing where clause with a websearch_to_tsquery match; order by ts_rank when q is present and keep created_at desc otherwise, inside the same sqlx::query_as! statement. Start Postgres in the sandbox, set DATABASE_URL, run sqlx db setup, then cargo fmt --check, cargo build and cargo test. Add tests/search.rs that inserts two articles through the API and asserts GET /api/articles?q= returns only the matching one. Start the server, curl the endpoint with and without q, and paste the responses in the pull request. Do not change the Realworld response shape, the JWT extractor or the existing migrations.",
  "resources": [
    {
      "url": "https://github.com/your-org/your-app"
    }
  ]
}
  1. Get an API key. In the console, open Settings, then API keys, and create a key for your org. Send it in the Authorization header as a Bearer token.
  2. Send the request. POST the body in request.json to https://api.phoenix.so/o/{org}/sessions, with your repository as the resources URL. The run starts in its own cloud sandbox.
  3. Read what came back. Follow the run on its live stream or wait for the run.completed webhook, then read the result and its trajectory.

What comes back

What comes back is a pull request with one new migration, migrations/5_article_search.sql, a q field on ListArticlesQuery and a longer where clause in src/http/articles/listing.rs, and a new tests/search.rs. The description shows the curl output for GET /api/articles?q=rust against the sandbox server, the cargo build line proving the macro accepted the new column, and the cargo test summary.

The repository's workflow applies the migrations with sqlx db setup, builds, runs cargo test and the Postman collection on the pull request, so a migration that fails to apply or a query the macro rejects shows up there. The Postman collection never sends q, so the new integration test and the pasted curl output are the evidence for the search path.

A reviewer checks that the generated column uses the english configuration their content actually needs, that the GIN index is on search and not on the three text columns, that websearch_to_tsquery rather than to_tsquery handles user input, and that the response shape is unchanged so the frontend keeps working. Then they run the migration on staging and watch the index build on a real-sized table.

Run it on every new list endpoint

A team adding search to several SQLx-backed services sends one POST /o/{org}/sessions per repository with this prompt and that repository in resources. Each run reads its own migrations and Cargo.toml, so a service on sqlx 0.7 with an offline .sqlx directory gets cargo sqlx prepare run as well, and each opens its own pull request on its own branch.

Send a request_id with each body so a retried request returns the run it already started instead of a second one. With a webhook_url on the API key, Rush POSTs run.completed when each turn ends, with the session_url to read the pull request from.

The use case this job follows, with its automation and trigger: Hand off a long task.

Why run it on Rush?

SQLx checks every query against a live database at compile time, so adding a column is not an edit you can make blind: the migration must apply, the database must be up and DATABASE_URL must point at it before cargo build passes. In a Rush sandbox the agent runs Postgres, sqlx db setup and cargo build itself and keeps every command and its output in the run's record, so a reviewer sees the macro accept the new search column rather than trusting a diff.

The proof is a response. The agent inserts articles through the real POST /api/articles route, queries /api/articles?q=… against the real server, and pastes what came back in the pull request. The same request body works unchanged on any Axum service a team runs, and runs for several repositories proceed at once, each in its own sandbox.

Frequently asked questions

Why a generated column instead of a trigger?

The article table already has an updated_at trigger from trigger_updated_at('article') in 4_article.sql. A stored generated column recomputes the tsvector on every write with no second trigger and no ordering question between them, and it is one line in a migration.

Will the build still work in CI?

Yes. .github/workflows/rust.yml already runs cargo install sqlx-cli and sqlx db setup against a Postgres service before cargo build, so the new migration applies there and the query_as! macro sees the search column. Nothing about the workflow changes.

Does it change the response shape?

No. MultipleArticlesBody still returns articles and articlesCount, and the tag, author, favorited, limit and offset filters keep working. q is one more optional query parameter; a request without it behaves as before.

What about search in languages other than English?

The migration uses the english text search configuration, which stems English words. For another language the agent changes the configuration name in the generated column and the two websearch_to_tsquery calls; the pull request notes where, and a mixed-language site would use simple instead.

More Axum jobs

The same job on other stacks

Keep reading