// Install and run Anserini quickly by downloading the latest fatjar from the official Maven Central repo instead of building from source. Use when users want fast setup, smoke tests, or command execution without Maven project compilation.
Use when onboarding to nuggetizer, ragnarok, rank_llm, or umbrela and you need development environment setup for one repo or several repos at once, including clone-if-needed, uv or pip installation paths, shared virtualenv reuse, and smoke tests.
Use when serving Anserini retrieval together with any subset of rank_llm, ragnarok, nuggetizer, or umbrela over HTTP, especially for local port planning, direct request payload compatibility, curl or jq pipelines, or sequencing retrieval, reranking, generation, nugget creation, nugget assignment, and passage judging from an Anserini fatjar RestServer.
Use when building, debugging, or reviewing CLI commands across nuggetizer, ragnarok, rank_llm, or umbrela and you need the shared castorini.cli.v1 envelope, common introspection commands, artifact shapes, or cross-repo CLI consistency rules.
Use when coordinating an end-to-end Castorini pipeline across rank_llm, ragnarok, nuggetizer, and umbrela, especially for stage handoffs, JSONL compatibility, retrieval-to-answer evaluation flow, or reproducing a multi-stage experiment.
Use when publishing nuggetizer, ragnarok, rank_llm, or umbrela to PyPI or TestPyPI and you need the release sequence for version bumps, build checks, twine validation, TestPyPI dry-runs, or final production publishing.
| name | anserini-fatjar |
| description | Install and run Anserini quickly by downloading the latest fatjar from the official Maven Central repo instead of building from source. Use when users want fast setup, smoke tests, or command execution without Maven project compilation. |
Use this skill to install and run Anserini quickly by downloading the latest fatjar from Maven Central into the current working directory.
Validation note:
anserini-1.7.1-fatjar.jarcacm-download reproduction, catalog lookups, Search CLI, and RestServerFor a reusable end-to-end check of the documented workflow, run:
SKILL_DIR="/path/to/anserini-fatjar"
bash "$SKILL_DIR/scripts/verify_fatjar_workflow.sh"
Run:
java -version
javac -version
wget --version
Require:
java and javacwget available on PATHjq available on PATH for JSON filtering commands belowStop immediately and warn the user if prerequisites are not satisfied. In particular:
java -version or javac -version does not report JDK 21, do not continue.wget is missing, do not continue with Maven Central download steps.jq is missing, continue only when JSON filtering is unnecessary or replace those examples with an equivalent plain-text inspection.From the skill directory, run:
SKILL_DIR="/path/to/anserini-fatjar"
bash "$SKILL_DIR/scripts/fetch_latest_fatjar.sh"
This script probes Maven Central metadata at https://repo1.maven.org/maven2/io/anserini/anserini/maven-metadata.xml,
extracts the current release, and downloads anserini-<version>-fatjar.jar from the matching official artifact path.
At the moment, a direct example URL is:
wget https://repo1.maven.org/maven2/io/anserini/anserini/1.7.1/anserini-1.7.1-fatjar.jar
All subsequent commands assume the jar is available in the current working directory as anserini-<version>-fatjar.jar.
Resolve the filename dynamically before invoking Java:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
Run:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" \
io.anserini.reproduce.ReproduceFromDocumentCollection \
--download --index --verify --search --config cacm-download
Treat the cacm-download reproduction as proof the runtime is ready. This workflow downloads the
CACM collection, builds the index, verifies index statistics, runs retrieval, and checks the
expected regression metrics from the config bundled in the jar.
For BM25, the expected scores from cacm-download are:
MAP = 0.3123P30 = 0.1942If the command exits successfully and the BM25 verification matches those values, the fatjar is working correctly.
Run Anserini commands against the downloaded jar:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" <main-class> <args>
Keep all commands pinned to the same downloaded jar version unless the user asks to change versions.
To inspect the prebuilt indexes exposed by io.anserini.cli.PrebuiltIndexCatalog, run:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" io.anserini.cli.PrebuiltIndexCatalog --list
--list emits JSON in the current jar, so prefer pairing it with --filter and jq instead of
grepping raw output when you need to identify a specific index.
msmarco-v1-passage is a particularly common choice and should be called out when users ask about available prebuilt indexes or MS MARCO passage retrieval setup.
Recommended lookup for the standard MS MARCO V1 passage inverted index:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" \
io.anserini.cli.PrebuiltIndexCatalog \
--list --filter '^msmarco-v1-passage$' \
| jq '.[0] | {name, type, description, filename}'
Useful variants:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"; java -cp "$ANSERINI_JAR" io.anserini.cli.PrebuiltIndexCatalog --help
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"; java -cp "$ANSERINI_JAR" io.anserini.cli.PrebuiltIndexCatalog --list --filter 'msmarco.*passage' | jq '.[].name'
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"; java -cp "$ANSERINI_JAR" io.anserini.cli.PrebuiltIndexCatalog --type flat --list
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"; java -cp "$ANSERINI_JAR" io.anserini.cli.PrebuiltIndexCatalog --type inverted --list
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"; java -cp "$ANSERINI_JAR" io.anserini.cli.PrebuiltIndexCatalog --type impact --list
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"; java -cp "$ANSERINI_JAR" io.anserini.cli.PrebuiltIndexCatalog --type hnsw --list
To inspect the topics exposed by io.anserini.cli.TopicsCatalog, run:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" io.anserini.cli.TopicsCatalog --list
--list emits JSON in the current jar, so prefer pairing it with --filter and jq to locate
the exact symbol you need.
To print all topics for a specific set, run:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" io.anserini.cli.TopicsCatalog --get <set>
For the standard MS MARCO V1 passage queries that pair with the msmarco-v1-passage prebuilt
index, use msmarco-v1-passage-dev.
Recommended lookup:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" \
io.anserini.cli.TopicsCatalog \
--list --filter '^msmarco-v1-passage-dev$' \
| jq '.'
In the current fatjar, --list exposes the canonical set name as msmarco-v1-passage-dev.
Use --list first to discover the exact set name, then --get to inspect its contents:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" io.anserini.cli.TopicsCatalog --get msmarco-v1-passage-dev
Use io.anserini.cli.Search for ad hoc retrieval against either a local Lucene index path or a prebuilt index name.
In the current fatjar, exactly one of --json or --trec is required.
Recommended example using the local CACM index created by the smoke test:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" \
io.anserini.cli.Search \
--index indexes/lucene-inverted.cacm.download \
--query "computer programming" \
--json --hits 10
Interactive mode:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" \
io.anserini.cli.Search \
--index indexes/lucene-inverted.cacm.download \
--interactive --json
Useful output variants:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"; java -cp "$ANSERINI_JAR" io.anserini.cli.Search --index indexes/lucene-inverted.cacm.download --query "computer programming" --json --hits 10
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"; java -cp "$ANSERINI_JAR" io.anserini.cli.Search --index indexes/lucene-inverted.cacm.download --query "computer programming" --trec --hits 10
The msmarco-v1-passage prebuilt index remains a useful option, but it is a heavyweight example:
the catalog currently advertises a compressed download of about 2.17 GB. Prefer the local CACM
index for smoke tests and quick command validation.
Use io.anserini.api.RestServer to expose search and document lookup over HTTP.
Fatjar invocation:
ANSERINI_JAR="$(ls -1 anserini-*-fatjar.jar | sort -V | tail -n 1)"
java -cp "$ANSERINI_JAR" io.anserini.api.RestServer --host 127.0.0.1 --port 8080
On startup, the server prints the route shape:
GET /v1/{index}/search?query=...&hits=10
GET /v1/{index}/doc/{docid}
If {index} is a filesystem path instead of a prebuilt index name, URL-encode it before sending requests.
For the local CACM index created by the smoke test, this works:
curl -sS 'http://127.0.0.1:8080/v1/indexes%2Flucene-inverted.cacm.download/search?query=computer%20programming&hits=2'
curl -sS 'http://127.0.0.1:8080/v1/indexes%2Flucene-inverted.cacm.download/doc/CACM-1771'
If the server cannot bind to the default port 8080, inform the user explicitly before trying a
different port. Do not silently switch ports.
If working inside an Anserini checkout, the equivalent helper script is:
bin/run.sh io.anserini.api.RestServer --host 127.0.0.1 --port 8080
Sample requests against the local CACM index created by the smoke test:
curl "http://127.0.0.1:8080/v1/indexes%2Flucene-inverted.cacm.download/search?query=computer%20programming&hits=5"
curl "http://127.0.0.1:8080/v1/indexes%2Flucene-inverted.cacm.download/doc/CACM-1771"
The document fetch route is /doc/{docid}, not /documents/{docid}.
This REST workflow works well with local indexes created during the smoke test. In restricted
environments, binding a local port may require elevated permissions.
wget -qO- https://repo1.maven.org/maven2/io/anserini/anserini/maven-metadata.xml and confirm network access to Maven Central.https://repo1.maven.org/maven2/io/anserini/anserini/<version>/.ClassNotFoundException: confirm the jar filename and working directory, then recompute ANSERINI_JAR.wget missing: warn the user and stop before attempting the Maven download.8080: inform the user that the default port is blocked or already in use before proposing an alternate port.java and javac on PATH and confirm no older JDK is shadowing them.Treat setup as complete when all are true:
anserini-<version>-fatjar.jar exists locally.ReproduceFromDocumentCollection --download --index --verify --search --config cacm-download executes successfully.MAP = 0.3123 and P30 = 0.1942.java -cp ....