A good starting point when searching for Rust crates is blessed.rs It serves as an unofficial guide to the Rust ecosystem. Here is another set of crates that can help out anyone writing any decent sized Rust project.

I put this list together while going through Rust infrastructure projects and I am keeping it mainly as a personal reference when I have to go crate hunting.

Application and Configuration Dependencies

Dependency	What it does / where it is used
`cfg-if`	Conditional compilation helpers, used where platform or feature-specific code needs a single source-level branch.
`clap`	Parses the CLI arguments, subcommands, environment-backed options, and help text.
`clap_complete`	Generates shell completion scripts.
`indoc`	Keeps multiline help text, templates, and test data readable without unwanted indentation.
`exitcode`	Standardized process exit-code constants for CLI failures.
`colored`	Terminal color output for CLI diagnostics and user-facing output.
`semver`	Parses and compares component, protocol, or API versions.
`uuid`	Generates and serializes IDs, with v4 and v7 support enabled.
`dirs-next`	Optional OS-specific configuration/cache/home-directory discovery, used by Docker-related behavior.
`notify`	Watches files, primarily to reload configuration and related local state.
`glob`	Expands file patterns, used in configuration and file-oriented behavior.
`hostname`	Obtains the local host name for event metadata and host-oriented configuration.
`bytesize`	Parses and formats human-friendly byte quantities such as buffer limits.
`humantime`	Workspace-shared duration parsing/formatting support.
`chrono`	Timestamps and date/time conversions.
`chrono-tz`	IANA time-zone support for config parsing and time operations.
`url`	Parses and validates endpoint URLs in component configuration.
`percent-encoding`	URL/query escaping and decoding.
`encoding_rs`	Character-set conversion for codecs and configuration.
`toml`	TOML configuration parsing.
`serde_yaml`	YAML configuration parsing.
`serde_json`	JSON configuration, event payloads, schemas, HTTP payloads, and arbitrary JSON values. `preserve_order` is enabled, so map ordering survives where relevant.
`serde`	The base serialization framework used throughout component configs and events.
`serde_with`	More expressive `serde` adapters for common config representations.
`serde_path_to_error`	Reports the exact nested config path when deserialization fails.
`http-serde`	`serde` support for HTTP types.
`typetag`	Serializes/deserializes trait-object-like configuration types.
`dyn-clone`	Makes boxed trait objects clonable.
`derivative`	Generates customized `Debug`, `Clone`, and similar implementations.
`derive_more`	Generates small conversion and display implementations.
`strum`	Derives enum iteration, parsing, display, and related helpers.
`inventory`	Link-time registration of component definitions. This is one of the mechanisms that lets Vector discover component implementations without a central manually maintained list.
`enum_dispatch`	Generates dispatch from enums to trait implementations without dynamic dispatch in selected paths.
`arr_macro`	Builds fixed-size arrays in places where ordinary array initialization is awkward.
`const-str`	Workspace-shared compile-time string manipulation.
`convert_case`	Workspace-shared string case conversion, mainly useful to procedural macros and generated docs/schema naming.
`pastey`	Workspace-shared identifier concatenation for macro-generated code.

Async runtime, streams, and flow-control

Dependency	What it does / where it is used
`tokio`	The async runtime. Enable `full` for production code and uses tasks, channels, timers, networking, signals, process support, and synchronization.
`futures`	Core async traits and combinators.
`futures-util`	Stream, future, sink, and async utility extensions.
`async-stream`	Builds async `Stream`s with generator-like syntax.
`async-trait`	Lets traits expose async methods before native async trait ergonomics cover all needed patterns.
`tokio-stream`	Turns Tokio channels, listeners, timers, and watches into streams.
`tokio-util`	Tokio codecs, I/O helpers, network helpers, cancellation/time support, and `DelayQueue` style utilities.
`stream-cancel`	Makes streams cancellable when topology shutdown begins.
`pin-project`	Safe pin-projection for custom streams, futures, and sinks.
`tower`	Composable service middleware. Uses buffering, limits, retries, timeouts, balancing, discovery, and utility middleware. Its batch/driver machinery uses `tower::Service`
`tower-http`	HTTP-specific Tower middleware for tracing plus request/response compression and decompression.
`tracing-tower`	Connects Tower requests to tracing spans.
`console-subscriber`	Optional Tokio-console instrumentation, enabled by the `tokio-console` feature.
`crossbeam-utils`	Workspace-shared concurrency utilities used by member crates.
`dashmap`	Workspace-shared concurrent map implementation for member crates.
`arc-swap`	Atomic replacement of shared immutable state. Used by the optional EC2 metadata transform, where readers need cheap consistent snapshots.
`evmap`, `evmap-derive`	Optional eventually-consistent read-optimized maps for in-memory enrichment tables.
`thread_local`	Optional thread-local state used by the in-memory enrichment implementation.
`lru`	Bounded least-recently-used cache.
`governor`	Optional rate limiting, used by `throttle` and Windows Event Log ingestion.
`cuckoo-clock`	Optional expiration/clock structure for memory enrichment tables.
`bloomy`	Optional Bloom-filter-style membership support for enrichment and tag-cardinality control.
`hashbrown`	Optional high-performance hash map, used by tag-cardinality limiting.
`hash_hasher`, `seahash`	Fast hashing in performance-sensitive internal data structures.
`smallvec`	Stores a few elements inline to avoid allocations on common small collections.
`indexmap`	Insertion-ordered maps, especially useful for preserving user configuration and JSON object order.
`ordered-float`	Makes floating-point values usable in ordered keys and comparisons.
`itertools`	Iterator combinators beyond the standard library.
`rand`, `rand_distr`	Sampling, random IDs/choices, randomized behavior, and distributions.
`regex`	Pattern matching for transforms, filters, parsers, and configuration validation.
`nom`	Parser combinators, notably enabled for Nginx metrics and ClickHouse-specific parsing.
`csv`	CSV decoding/encoding support.
`quick-junit`	Emits JUnit test result data.
`tempfile`	Creates temporary files and directories for runtime features and tests.
`libc`, `nix`	Unix syscalls, sockets, signals, file descriptors, and resource limits.
`socket2`	Lower-level socket setup where `std::net` is insufficient.
`heim`	Host system metrics, used by the `host_metrics` source.
`sysinfo`	System inspection support.
`procfs`	Linux `/proc` parsing.
`tikv-jemallocator`	Optional Unix allocator selected by default on supported builds. It reduces allocator contention in allocation-heavy workloads.

HTTP, TLS and RPC

Dependency	What it does / where it is used
`http`	HTTP 0.2 request, response, headers, methods, and status types.
`http-1`	Renamed `http` 1.x, kept alongside 0.2.
`http-body`	HTTP 0.4 body abstraction.
`http-body-1`	Renamed HTTP body 1.x abstraction.
`http-body-util`	Utilities for collecting and transforming HTTP 1.x bodies.
`hyper`	HTTP 0.14 client/server implementation.
`hyper-1`	Renamed Hyper 1.x client.
`hyper-util`	Hyper 1.x Tokio/client adapters, including legacy client compatibility.
`axum`	HTTP routing and server-layer support
`warp`	Existing HTTP server/filter implementation used by parts of the application.
`headers`, `headers-1`	Typed HTTP header support for the two HTTP API versions.
`h2`	Optional HTTP/2 protocol support, used by GCP Pub/Sub ingestion.
`reqwest`	HTTP client
`reqwest_13`	A renamed Reqwest 0.13 client, used where a newer HTTP stack is required alongside the 0.11 dependency.
`hyper-proxy`	Proxy-aware HTTP client support.
`openssl`	Vendored OpenSSL TLS and crypto implementation.
`openssl-probe`	Finds system CA certificate locations where needed.
`tokio-openssl`	Async Tokio integration for OpenSSL streams.
`hyper-openssl`, `hyper-openssl-1`	TLS connectors for the Hyper 0.14 and Hyper 1.x clients.
`flate2`	gzip/deflate compression using `zlib-rs`.
`zstd`	Zstandard compression. Also used for compression dictionary generation in tests.
`snap`	Snappy compression support.
`async-compression`	Optional asynchronous gzip and zstd streams, used by S3 input and file output.
`base64`	Optional binary-to-text encoding used by protocols and cloud integrations.
`hex`	Optional hexadecimal encoding, used by OpenTelemetry ingestion.
`md-5`	Optional MD5 hashing for S3 object/checksum behavior.
`prost`	Optional protobuf runtime used by Vector API, Vector protocol, Datadog, OpenTelemetry, Prometheus, GCP Pub/Sub, and dnstap.
`prost-types`	Standard protobuf well-known message types.
`prost-reflect`	Optional runtime protobuf reflection, used by Datadog metrics.
`prost-build`	Optional build-time Rust code generation from `.proto` files.
`tonic`	Optional gRPC client/server implementation for Vector API, Vector source/sink, GCP Pub/Sub, and other protobuf endpoints.
`tonic-build`	Optional build-time gRPC/protobuf code generation.
`tonic-health`	gRPC health service implementation.
`tonic-reflection`	gRPC reflection server support.
`protobuf`	Alternative protobuf runtime, used by Datadog metrics support.
`openssl-src`	Build dependency that compiles/configures the OpenSSL source used by the vendored TLS setup.

Event model, codecs, and VRL

Dependency	What it does / where it is used
`mlua`	Optional vendored Lua 5.4 runtime used by the `lua` transform.
`rmp-serde`	Optional MessagePack serialization/deserialization through Serde. Used by Fluent and Datadog trace paths.
`rmpv`	Optional dynamic MessagePack values, also used by Fluent and Datadog trace handling.
`serde_bytes`	Optional efficient serialization of binary fields without expanding them into integer arrays.
`arrow`	Optional Apache Arrow columnar-memory format support for Arrow and Parquet codecs, ClickHouse, and Databricks Zerobus.
`parquet`	Optional Apache Parquet writer/reader support.
`datadog-agent-metrics-v3`	Optional Datadog metrics-v3 columnar codec.
`databricks-zerobus-ingest-sdk`	Optional Arrow Flight ingest client for the Databricks Zerobus sink.
`roaring`	Optional compressed bitmap support for Splunk HEC source bookkeeping.
`strip-ansi-escapes`	Removes terminal ANSI escapes from text where Vector needs clean content.
`syslog`	Optional syslog protocol support for the Papertrail sink.

Brokers and Messaging Systems

Dependency	What it enables
`lapin`	AMQP client for the AMQP source and sink.
`async-rs`	Async AMQP protocol/runtime support paired with `lapin`.
`deadpool`	Managed async connection pool for the AMQP sink.
`rdkafka`	Kafka client for Kafka sources and sinks. Enables static curl, TLS, compression, and Tokio support.
`async-nats`	NATS and JetStream client for NATS sources and sinks.
`nkeys`	NATS NKey authentication.
`pulsar`	Apache Pulsar source and sink client.
`redis`	Redis source and sink client, including connection manager and Sentinel support.
`rumqttc`	MQTT source and sink client using Rustls.
`mongodb`	MongoDB metrics source.
`postgres-openssl`	OpenSSL TLS connector for PostgreSQL telemetry collection.
`tokio-postgres`	PostgreSQL metrics source.
`sqlx`	PostgreSQL sink and the MySQL-enabled Doris sink path.
`databend-client`	Databend sink client.
`greptimedb-ingester`	GreptimeDB logs and metrics sinks.
`maxminddb`	GeoIP and MMDB enrichment tables.
`ipnet`	CIDR/IP-network handling for TCP source utilities.

Platform-specific Dependencies

Platform	Dependencies	Role
Unix	`libc`, `nix`	Signals, sockets, file descriptors, and resource handling.
Unix	`tikv-jemallocator`	Optional default allocator on supported Unix builds.
Unix	`dnsmsg-parser`, `dnstap-parser`	dnstap source implementation.
Linux	`procfs`	`/proc` metrics and system data.
Linux	`antithesis-instrumentation`	Optional Antithesis fault-testing instrumentation.
Windows	`windows-sys`	Basic Windows FFI for foundations and threading.
Windows	`windows-service`	Windows service integration.
Windows	`windows`	Optional higher-level Windows API bindings for the Windows Event Log source.
Windows	`quick-xml`	Parses Windows Event Log XML records.
Windows	`rdkafka` with `cmake_build`	Builds Kafka native dependencies on Windows.

Build-only Dependencies

Dependency	What it does
`prost-build`	Generates Rust types from protobuf files when the enabled feature requires them.
`tonic-build`	Generates gRPC service/client code from protobuf definitions.
`openssl-src`	Supplies the vendored OpenSSL build used by the application’s TLS configuration.

Test and Benchmark Dependencies

Dependency	What it does
`approx`	Floating-point assertions with tolerances.
`assert_cmd`	Runs a CLI as a child process and asserts stdout, stderr, and exit status.
`criterion`	Statistical benchmarks, including async Tokio benchmarks and HTML reports.
`mock_instant`	Controllable clock for deterministic timing tests.
`parking_lot`	Fast locks in test scenarios.
`proptest`, `proptest-derive`	Property-based test generation and derived arbitrary values.
`quickcheck`	Another property-based test framework.
`rstest`	Parameterized test macros.
`serial_test`	Serializes tests that mutate global/shared state.
`similar-asserts`	Better diff output when assertions fail.
`test-generator`	Generates repeated tests from declarative inputs.
`tokio-test`	Tokio-specific async testing helpers.
`tower-test`	Mocks and test helpers for Tower services.
`wiremock`	Mock HTTP server for integration-style client tests.
`zstd` with `zdict_builder`	Trains compression dictionaries used to generate test fixtures.