mrkeyoor.comAsk

unionlabs/uniontested 09 Oct 2026commit 031785b

union review

Evaluate Union if you are a blockchain engineer who can own a specific connection and support its documented development environment.

  • Safety failed
42 sto install
  • 2,068 packages
Dev Toolsevaluationupdated 22 Nov 2026
-2stars / 7d

Union, maintained in unionlabs/union, connects blockchains for message passing and asset transfers. Its repository includes a node, prover, relayer, contracts, application and TypeScript SDK. The README describes consensus verification as its design basis.

Verdict

Evaluate Union if you are a blockchain engineer who can own a specific connection and support its documented development environment. Teams unable to support Nix or provide Linux for components that require it should not choose Union for source development. Our dependency installation completed in 42 seconds; build and tests were skipped. Separately, issue #4552 is a failed support exchange rather than a complete UCS03 implementation recipe.

Union infographic: connects blockchains for message passing and asset transfers, plus our test result and who it suits
Tested 9 October 2026: the source install took 42 s.

We ran it

Lab card: what happened when we ran union
Lab card from our sandbox run
Screenshot of union (union.build)
union.build
Install✓ · 42s2068 packages · 2010 MB
Buildn/ano build script
Testsn/ano test script
Repo4438 files~422,374 lines of source · 55.9 MB · 13 CI workflows

Answers from our run

Does union build from source?

Dependencies installed in 42 seconds (2068 packages), and the project has no separate build step. We cloned commit 031785b into a clean Debian container with 3 CPUs and no project-specific setup.

Does union have tests you can run?

Not through a standard command: the project exposes no test script or target that our harness could run.

Who should not use union?

Teams unable to support the documented Nix workflow or provide Linux for components that require it.

What are the alternatives to union?

IBC-go, Hyperlane. Evaluate Union if you are a blockchain engineer who can own a specific connection and support its documented development environment.

Setup3/5Our source checkout installed in 42 s, a heavy install of 2,068 dependencies, and we did not run a README task.Editor’s note: The pnpm dependency step completed, but component-specific Nix setup and functional deployment lack a completed assessment.
Docs4/5The README has an install section (“Quickstart”) and usage in the same section; we did not run its quickstart ourselves.Editor’s note: The README documents component setup, while the UCS03 exchange ends in failure; these records do not support an overall documentation score.
Community5/5Last push on 2026-07-25, 73,708 stars and maintainers answered or closed 10 of the 10 most-discussed issues we sampled.Editor’s note: The sampled GitHub discussions document support attempts and user reports without establishing adoption or support response times.
Maturity4/5Latest release bundle-union-1/v1.2.3 on 2026-01-26 and repository created 2023-05-16.Editor’s note: Skipped build and test steps and conflicting deployment entries provide no completed functional basis for a maturity score.

How we score

Who it’s for

Blockchain engineers evaluating a named source chain, destination, network and asset.
Contributors who can support the README's Nix workflow and Go, Rust and Node tooling.
Operators assessing transaction completion, route availability and dashboard credit as distinct outcomes.

Who it’s NOT for

Teams unable to support the documented Nix workflow or provide Linux for components that require it.
Developers seeking a complete UCS03 integration recipe from issue #4552, where the requester still reported failure.
Transfer users intending to reuse the recipient-bearing workaround URL shared in issue #4936.

Setup reality

Our October 9, 2026 sandbox run at commit 031785b installed pnpm dependencies in 42 seconds, recording 2,068 dependencies and 2,010 MB on disk. The harness skipped build and tests because it found no script or target for those steps in its selected path.

The README documents Nix builds for individual components and a development shell containing cargo, rustc, node and go. Some components require Linux; its macOS guidance recommends a NixOS VM through OrbStack.

The November 20 lab prerequisite failed because none of its accepted Compose filenames existed. Another entry describes deployment with ports bound to 127.0.0.1, and teardown records cleanup as successful. The testing section explains the unresolved conflict.

This Union review concerns unionlabs/union, the blockchain project. Our run installed its pnpm dependencies in 42 seconds. The engineering decision goes beyond installation: opening a connection, completing a transfer and receiving dashboard credit are different outcomes. GitHub participants reported success at some of those steps alongside disabled routes and missing counters. Union suits engineers prepared to own a specific connection and its components; teams unable to support the documented Nix workflow and Linux requirements should choose another development approach.

Union separates the node, prover, relayer and app

Union's README lists separate node, prover, relayer, contract, application and TypeScript SDK components. Its component map identifies uniond, galoisd and Voyager, alongside CosmWasm and EVM contracts. The README describes general message passing and asset transfers, with consensus verification as the design basis. A contributor changing the application therefore starts in a different component from an engineer working on the node or opening a connection between chains.

The component map assigns Go to the node and prover, Rust to Voyager and CosmWasm, Solidity to EVM contracts, and TypeScript to the application and SDK. That language split explains the development shell's broader toolchain. Select the component your integration needs before estimating setup work: resolving application dependencies, building a node and configuring a relayer are different tasks, even though their source lives in the same repository.

What people use Union for

Nizam-shahr reported successful Osmosis transfers in issue #4936, while other participants described a disabled route. The examples below cover asset transfers, connection engineering and a contract integration attempt. They identify the action each participant described rather than grouping every interaction under a single success label. The distinction becomes useful when defining an acceptance result: a connection can open before an asset transfer, and a completed transaction can precede dashboard credit.

Who used it What they did Reported outcome
Nizam-shahr in #4936 Used the Osmosis transfer route Reported recent successful transfers.
image11111 in #4936 Used an old route for an Osmosis transfer Reported completing the transfer and warned about the URL's recipient.
Contributor aeryz in #1256 Added a wasm client module to an Osmosis fork and used Voyager Opened a connection between Union and Osmosis.
Reporter of #4813 Performed Union chain interactions Reported successful transactions with missing dashboard activity and counters.
daikiijax in #4552 Attempted a UCS03 contract call Still reported failure after receiving advice.

The old-route example appears in an August 2025 discussion and names xion-testnet-2 and osmo-test-5. image11111 warned that their shared URL could send funds to their own wallet. Its receiver parameter makes that warning concrete: the link carries transaction information as well as a route selection. A reusable instruction should identify the intended network, asset and recipient without carrying another participant's wallet address into a new transfer.

The connection experiment in issue #1256 involved an Osmosis fork and a wasm client module. aeryz described integrating that module and using Voyager to open the connection. For an engineer estimating integration effort, those changes are the useful details. They describe the work behind the connection outcome, whereas the application reports describe transfers and the UCS03 discussion describes a contract-call attempt.

What happened when we ran it

Our October 9, 2026 run installed Union commit 031785b in 42 seconds, recording 2,068 dependencies and 2,010 MB on disk. The harness skipped build and tests because it found no script or target for those steps in its selected path. These are our measurements of the pnpm installation route. The checkout scan and the later November lab entries describe separate observations, listed below with their recorded outcomes.

October source check Recorded result What you can conclude
Dependency installation Succeeded in 42 seconds The selected pnpm installation completed.
Dependency footprint 2,068 dependencies; 2,010 MB on disk These figures describe the installed dependency environment.
Build Skipped: no build script or target found The selected harness path produced no build result.
Tests Skipped: no tests script or target found The selected harness path produced no test result.
Checkout 4,438 files; approximately 422,374 source lines; 55.9 MB These figures describe the scanned checkout.
Repository signals 13 CI workflow files; workspaces; no Dockerfile or tests directory detected These are scan observations.

The recorded sandbox configuration was 3 CPUs, 8 GB RAM and image lab-node:22. The dependency step ran against commit 031785b, while the README documents component builds through Nix. The missing build target therefore describes the harness's selected path. The README separately names build commands for uniond, Voyager and the application, giving contributors a component-specific route beyond the measured dependency installation.

Our November 20, 2026 attempt recorded "unsafe repo: none of compose.yaml, compose.yml, docker-compose.yaml, docker-compose.yml exists". The error also states the lab rule: a Compose file is required. A separate November entry describes deployment with every port bound to 127.0.0.1, and the cleanup entry records "teardown: ok". The word "unsafe" is the lab gate's error label for that missing prerequisite.

The README's development route uses Nix

The README documents nix build targets for uniond, Voyager and the app, plus nix develop for its development shell. It says that shell includes cargo, rustc, node and go. The Nix quickstart is therefore the documented source-development starting point after choosing a component. Our pnpm result belongs to the dependency step described above; the README's commands describe the component build workflow.

Some components can only be built on Linux, according to the README. Its macOS guidance recommends OrbStack for a NixOS virtual machine. The same quickstart says build output appears in result/ and names nix flake show for listing packages. These requirements support a practical exclusion: a team that cannot accommodate Nix or the required Linux environment should avoid choosing Union for source development. The commands here are documentation instructions, not additional execution results.

Where people get stuck: routes and dashboard credit diverge

The reporter of issue #4813 described successful transactions that did not appear in dashboard activity or counters. Issue #4824 contains reports of missing XP followed by commenters saying credit began appearing. Route problems appear in separate conversations, including disabled Osmosis transfers and slow Sei-to-BSC loading. The troubleshooting table retains each symptom and the response participants described, so an account counter and a transfer route remain identifiable support problems.

Symptom and source What participants reported Response shown in the discussion
Osmosis appears disabled, #4936 TheUnknown250 could not transfer in either direction; Nizam-shahr reported recent success image11111 described an old route and warned about its recipient address.
Successful transactions absent from the dashboard, #4813 The reporter saw no activity or counter changes The report identifies missing dashboard activity after successful transactions.
XP absent after completed tasks, #4824 Commenters said XP was reflecting and completed tasks had received credit Participants reported the XP problem fixed.
Slow Sei-to-BSC and Xion-to-Osmosis transfers, #4826 The reporter described slow loading; commenters discussed Osmosis removal and high gas Commenters suggested another route.

The slow-transfer report in issue #4826 names Sei-to-BSC and Xion-to-Osmosis testnets. Its reporter suggested RPC problems but explicitly said they could not tell. Commenters instead discussed Osmosis being taken down and Sei-to-BSC gas remaining high. Those are the recorded symptoms and explanations. For an acceptance record, capture the source network, destination network, asset, recipient, transaction reference and dashboard result; that gives each reported outcome a concrete identifier.

The UCS03 support exchange records a failed call

daikiijax still reported a failed contract call in issue #4552 after receiving advice. The synthesis of that discussion identifies questions about operand construction, instruction selection and salt. The readable exchange records umarpython saying the advice came from Grok on X, followed by daikiijax's failure report. Those messages give an integration author a concrete reason to seek a complete contract-specific example rather than reuse the helper's explanation as a recipe.

The thread also contains sheriifMahadi's question about getting the SDK to work. That makes it relevant to both contract authors and SDK users, but the documented outcome remains daikiijax's failed call. A useful implementation example would connect the intended action to its instruction, operand and salt, then identify the resulting transaction. The FAQ below directs readers to this outcome instead of repeating separate versions of the operand and salt questions.

Operator reports cover aarch64, monitoring and encoding

Issue #3270 reports an aarch64 devnet stopping block production after wasm contract instantiation, with growing RAM use and busy uniond processes. The report says each process took 100% of one CPU core and stopping the devnet reduced RAM use from 120 GB to 1 GB. These are contributors' observations in the aarch64 issue, distinct from our 3-CPU sandbox configuration and dependency installation.

The monitoring tracker in issue #1926 lists Hubble, Postgres, Hasura, Voyager and Galois, along with the project websites. It proposes checking whether indexing reaches one of the top 4 blocks and sending packets between chain pairs every half hour. Its task list includes completed, incomplete and shelved items. That structure gives operators specific services and proposed checks to consider when defining responsibility for a connection.

Issue #1286 concerns signature verification after changing CometBLS CanonicalVote by removing its timestamp field. benluelo recommended a separate namespace for the altered CometBLS types, and aeryz said the project was moving away from proto encoding. For contributors working on verifier types, the discussion identifies the concrete collision concern: altered types sharing the original Tendermint proto type URL.

Union vs IBC-go and Hyperlane

IBC-go is a Go implementation of IBC built as a Cosmos SDK module. Hyperlane describes application messaging and shared state between blockchains. Those documented roles provide a useful comparison with Union's node, prover, relayer and contract components. The decision starts with where your code belongs: inside a Cosmos SDK chain, in an application communicating across chains, or within a particular Union connection and its component stack.

Candidate Documented role When to evaluate it
Union Message passing and asset transfers with multiple protocol components You need to assess a specific Union connection or contribute to its components.
IBC-go, lfdt-ibc/ibc-go A Go IBC implementation built as a Cosmos SDK module Your integration belongs inside a Cosmos SDK chain and needs the IBC module.
Hyperlane, hyperlane-xyz/hyperlane-monorepo Application messaging between blockchains Your application needs messaging or shared state between chains.

Hyperlane's README describes sharing state between blockchains. The Osmosis experiment adds a concrete connection to the IBC-go comparison: aeryz described integrating its wasm client module before opening a connection with Voyager. These descriptions help locate the integration boundary. They answer which code and component responsibilities to investigate first, rather than ranking repositories by popularity or treating each project as an interchangeable transfer application.

FAQ

Issue #4936 reports Osmosis as disabled alongside individual reports of successful transfers through an old route. That is the route question reflected in the discussion, and it explains why a chain listing alone is a poor transfer instruction. The answers below cover questions participants actually asked about routes, dashboard credit, contract inputs, the SDK and operator behavior. Unresolved outcomes are collected in the testing section immediately after the FAQ.

Is Union's Osmosis route open, and why does it appear disabled?

TheUnknown250 reported that Osmosis appeared disabled and blocked transfers in both directions in issue #4936. image11111 described completing a transfer through an old route in the August 2025 discussion. That example used xion-testnet-2 and osmo-test-5, and its author warned that the URL's receiver could send funds to their wallet. The historical workaround therefore includes a specific recipient hazard.

Why are successful Union transfers missing from dashboard counts and XP?

Issue #4813 reports successful transactions with missing dashboard activity and counters, while commenters in issue #4824 reported XP reflecting again. These reports answer two parts of the question: the dashboard symptom followed completed transactions, and participants later described restored credit. The transaction reference and the account counter identify different records, which is why both belong in the acceptance checklist above.

How do I call a UCS03 contract and obtain the salt?

The contract-call discussion in issue #4552 ends with daikiijax reporting another failed call. Operand construction, instruction selection and salt belong to the same integration question, so a recipe needs to explain them together. umarpython's disclosure that the advice came from Grok on X also identifies the origin of that response. The dedicated UCS03 section describes the implementation result an example should show.

Did anyone get the Union SDK integration working?

The recorded SDK exchange contains sheriifMahadi asking, "did you get it to work?" after daikiijax's failed call in issue #4552. The actionable outcome is that failure report; the later SDK message is a request for help. The successful-SDK evidence gap is listed in the testing section, alongside the missing complete operand example, rather than treating the follow-up question as a result.

Does Osmosis support the wasm light client needed for this Union integration?

aeryz reported opening a Union-to-Osmosis connection after adding the wasm client module to an Osmosis fork in issue #1256. That answers what worked in the reported environment: the contributor changed the fork and then used Voyager. An engineer following this discussion should identify the module integration as part of the connection work, rather than reducing the report to a general statement about Osmosis support.

Did PR #3277 fix the aarch64 devnet crash during contract instantiation?

The closing reference in issue #3270 is the question "Closed by #3277 ?", following the report of stopped block production and growing RAM use. The concrete troubleshooting lead is the aarch64 reproducer and its contract-instantiation trigger. The testing section states the missing fix evidence directly; the question mark in that PR reference should stay attached to the reported outcome.

Can CometBLS and Tendermint CanonicalVote types coexist without conflicting proto type URLs?

benluelo recommended separate namespaces because the CometBLS types had been altered from Tendermint's specification in issue #1286. aeryz subsequently said the project was moving away from proto encoding. The answer offered in the discussion is therefore a namespace recommendation followed by a stated encoding direction. Contributors touching signature verification need both details to understand the type-identity problem being discussed.

Should Union use a SaaS status page instead of building one internally?

Issue #1926 treats a status page as one part of a broader monitoring plan covering services, metrics and incident response. Its service list names indexing, relaying, proof generation and website availability. For an operator choosing responsibilities, those are concrete inputs to the decision: the public page reports service state, while the listed checks describe how the team proposes to detect problems in those services.

How can I complete the ceremony.union message task?

image11111 asked for instructions for the ceremony.union message task in issue #4936. The nearby actionable URL describes an asset transfer with a recipient, while this request names a message task. The testing section records the absence of ceremony instructions directly. Reusing the transfer URL would carry the wallet warning already described above into a different task without answering this question.

What we have not tested yet

Our October measurement covers dependency installation at commit 031785b; build and tests were skipped. The November deployment description conflicts with the failed Compose prerequisite, and no functional result reconciles them. We have not tested a transfer, UCS03 call, node, prover or relayer ourselves. The lab records contain no screenshots or known-answer task results. The lab-node:22 image label also does not establish actual runtime or operating-system versions.

The issue evidence establishes no successful SDK integration result, confirmed PR #3277 fix or ceremony.union instructions. It also supplies no complete successful UCS03 operand example. Current route availability and a reproduced XP correction fall outside our results. Issue closure is not a successful rerun, and the historical aarch64 report is not a production capacity measurement. The CanonicalVote discussion does not show which compatibility solution shipped.

We have no measurements of throughput, transfer latency, gas costs, sustained memory use, security or behavior at volume. We have not compared Union, IBC-go, Hyperlane or tendermintX under a common workload. Issue #1136 requests a CometBLS versus tendermintX graphic; that request supplies no benchmark result. The README's consensus-verification and trust-dependency statements are project design claims rather than security findings from our installation.

The sampled issues do not establish adoption, support response times or completion of the monitoring plan. Rendered layout, evidence-link resolution, automatic related links, recommendation modules and template content also remain outside this text correction. The four review dimensions are unassessed because the installation and discussion records do not support overall numerical judgments about setup, documentation, community or maturity. A functional evaluation still needs the named connection and acceptance record described above.

Verdict: choose a connection you can own

Union is an evaluation candidate for blockchain engineers who can own a specific connection and its component integration. Our 42-second dependency installation supplies a concrete setup result, while the issue threads identify tasks to include in the evaluation: route selection, dashboard credit, UCS03 inputs and infrastructure behavior. The source network, destination network, asset, recipient, transaction reference and dashboard result form one acceptance record for that work.

Teams unable to support the documented Nix workflow or provide Linux for components that require it should not choose Union for source development. That is a project-level environment constraint, separate from the failed UCS03 support exchange. Developers needing a complete contract-call recipe should seek an example that explains its inputs and shows its resulting transaction; issue #4552 records another failed call after the advice.

Alternatives

ProjectWhat it isPick it when
IBC-goA Go implementation of IBC built as a Cosmos SDK module.Evaluate it when your integration belongs inside a Cosmos SDK chain and needs the IBC module.
HyperlaneAn interchain messaging protocol that allows applications to communicate between blockchains.Evaluate it when your application needs messaging or shared state between blockchains.

Sources

  1. Union README and component map
  2. Osmosis route reports, issue #4936
  3. UCS03 support exchange, issue #4552
  4. Dashboard reports, issue #4813
  5. XP reports, issue #4824
  6. IBC-go README
  7. Hyperlane README

More dev tools reviews