rtk-ai · mariuszs · Apr 16, 2026 · Apr 17, 2026 · Apr 17, 2026 · Apr 17, 2026
diff --git a/.claude/rules/search-strategy.md b/.claude/rules/search-strategy.md
@@ -44,6 +44,7 @@ src/
 │   ├── js/                    ← npm, pnpm, vitest, lint, tsc, next, prettier, playwright, prisma
 │   ├── python/                ← ruff, pytest, mypy, pip
 │   ├── go/                    ← go, golangci-lint
+│   ├── jvm/                   ← mvn/mvnw/mvnd, gradlew (test, compile, checkstyle, dependency:tree)
 │   ├── dotnet/                ← dotnet, binlog, trx, format_report
 │   ├── cloud/                 ← aws, container (docker/kubectl), curl, wget, psql
 │   ├── system/                ← ls, tree, read, grep, find, wc, env, json, log, deps, summary, format, local_llm

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -240,6 +240,14 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Features
+
+* **mvn:** add Maven (Java) filter module — test, compile, checkstyle:check, dependency:tree ([#1089](https://github.com/rtk-ai/rtk/pull/1089))
+* **mvn:** enrich `mvn test` / `verify` / `integration-test` output with structured failure details read from `target/surefire-reports/TEST-*.xml` and `target/failsafe-reports/*.xml`. Stack traces are segmented on `Caused by:` with framework frames collapsed; the root-cause segment is always preserved.
+* **mvn:** autodetect application package from `pom.xml` `<groupId>` (with `<parent>/<groupId>` fallback) for framework-frame classification.
+* **mvn:** red-flag heuristic — `no tests run` with no fresh XML reports emits a diagnostic pointing at surefire misconfiguration.
+* **mvn:** signal-aware multi-goal filtering — filters every goal in a chain (`clean test-compile checkstyle:check`, `clean verify`, `clean install`) instead of only the first; always preserves BUILD SUCCESS/FAILURE, `[ERROR]`s, test counts, and checkstyle violations; auto-drops `-q`; enriches test failures from surefire/failsafe XML when the chain includes test-running goals.
+
 ### Bug Fixes
 
 * **git:** remove `-u` short alias from `--ultra-compact` to fix `git push -u` upstream tracking ([#1086](https://github.com/rtk-ai/rtk/issues/1086))

diff --git a/CLAUDE.md b/CLAUDE.md
@@ -75,7 +75,7 @@ For the full architecture, component details, and module development patterns, s
 
 Module responsibilities are documented in each folder's `README.md` and each file's `//!` doc header. Browse `src/cmds/*/` to discover available filters.
 
-Supported ecosystems: git/gh/gt, cargo, go/golangci-lint, npm/pnpm/npx, ruff/pytest/pip/mypy, rspec/rubocop/rake, dotnet, playwright/vitest/jest, docker/kubectl/aws.
+Supported ecosystems: git/gh/gt, cargo, go/golangci-lint, npm/pnpm/npx, ruff/pytest/pip/mypy, rspec/rubocop/rake, dotnet, mvn/mvnw, playwright/vitest/jest, docker/kubectl/aws.
 
 ### Proxy Mode
 

diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -41,6 +41,8 @@ libc = "0.2"
 toml = "0.8"
 
 [dev-dependencies]
+filetime = "0.2"
+insta = "1"
 
 [profile.release]
 opt-level = 3

diff --git a/README.md b/README.md
@@ -48,6 +48,7 @@ rtk filters and compresses command outputs before they reach your LLM context. S
 | `git log` | 5x | 2,500 | 500 | -80% |
 | `git add/commit/push` | 8x | 1,600 | 120 | -92% |
 | `cargo test` / `npm test` | 5x | 25,000 | 2,500 | -90% |
+| `mvn test` | 3x | 30,000 | 300 | -99% |
 | `ruff check` | 3x | 3,000 | 600 | -80% |
 | `pytest` | 4x | 8,000 | 800 | -90% |
 | `go test` | 3x | 6,000 | 600 | -90% |
@@ -181,6 +182,12 @@ rtk go test                     # Go tests (NDJSON, -90%)
 rtk cargo test                  # Cargo tests (-90%)
 rtk rake test                   # Ruby minitest (-90%)
 rtk rspec                       # RSpec tests (JSON, -60%+)
+rtk mvn test                    # Maven tests (-99%)
+rtk mvn verify                  # Maven verify — surefire + failsafe XML enrichment
+rtk mvn clean                   # Maven clean — one-line summary (-95%)
+rtk mvn clean test-compile checkstyle:check  # Multi-goal: per-goal filter, BUILD signal always preserved
+rtk mvn clean verify            # Multi-goal: XML enrichment applies; -q auto-dropped
+rtk mvnd test                   # Maven Daemon tests (same filter, same savings)
 rtk err <cmd>                   # Filter errors only from any command
 rtk test <cmd>                  # Generic test wrapper - failures only (-90%)
 ```
@@ -197,6 +204,8 @@ rtk cargo clippy                # Cargo clippy (-80%)
 rtk ruff check                  # Python linting (JSON, -80%)
 rtk golangci-lint run           # Go linting (JSON, -85%)
 rtk rubocop                     # Ruby linting (JSON, -60%+)
+rtk mvn build                   # Maven build (-90%)
+rtk mvn dependency:tree         # Maven dependency tree (-60%+)
 ```
 
 ### Package Managers

diff --git a/docs/contributing/TECHNICAL.md b/docs/contributing/TECHNICAL.md
@@ -299,7 +299,7 @@ Start here, then drill down into each README for file-level details.
 | [`core/`](../src/core/README.md) | Shared infrastructure | Tracking DB schema, config system, tee recovery, TOML filter engine, utility functions |
 | [`hooks/`](../src/hooks/README.md) | Hook system | Installation flow (`rtk init`), integrity verification, rewrite command, trust model |
 | [`analytics/`](../src/analytics/README.md) | Token savings analytics | `rtk gain` dashboard, Claude Code economics, ccusage parsing |
-| [`cmds/`](../src/cmds/README.md) | **Command filters (9 ecosystems)** | Common filter pattern, cross-command routing, token savings table, **links to each ecosystem** |
+| [`cmds/`](../src/cmds/README.md) | **Command filters (10 ecosystems)** | Common filter pattern, cross-command routing, token savings table, **links to each ecosystem** |
 | [`discover/`](../src/discover/README.md) | History analysis + rewrite registry | Rewrite patterns, session providers, compound command splitting |
 | [`learn/`](../src/learn/README.md) | CLI correction detection | Error classification, correction pair detection, rule generation |
 | [`parser/`](../src/parser/README.md) | Parser infrastructure | Canonical types (TestResult, LintResult, etc.), 3-tier format modes, migration guide |

diff --git a/src/cmds/README.md b/src/cmds/README.md
@@ -30,6 +30,7 @@ Each subdirectory has its own README with file descriptions, parsing strategies,
 - **[`js/`](js/README.md)** — npm, pnpm, vitest, lint, tsc, next, prettier, playwright, prisma — Package manager auto-detection, lint routing, cross-deps with python
 - **[`python/`](python/README.md)** — ruff, pytest, mypy, pip — JSON check vs text format, state machine parsing, uv auto-detection
 - **[`go/`](go/README.md)** — go test/build/vet, golangci-lint — NDJSON streaming, Go sub-enum pattern
+- **[`jvm/`](jvm/README.md)** — mvn/mvnw/mvnd, gradlew — State-machine test parser, Surefire/Failsafe XML enrichment, checkstyle compaction, dependency:tree collapsing
 - **[`dotnet/`](dotnet/README.md)** — dotnet, binlog, trx, format_report — DotnetCommands sub-enum, internal helper modules
 - **[`cloud/`](cloud/README.md)** — aws, docker/kubectl, curl, wget, psql — Docker/Kubectl sub-enums, JSON forced output
 - **[`system/`](system/README.md)** — ls, tree, read, grep, find, wc, env, json, log, deps, summary, format, smart — format_cmd routing, filter levels, language detection

diff --git a/src/cmds/jvm/README.md b/src/cmds/jvm/README.md
@@ -0,0 +1,63 @@
+# JVM Ecosystem (Maven, Gradle)
+
+> Part of [`src/cmds/`](../README.md) — see also [docs/contributing/TECHNICAL.md](../../../docs/contributing/TECHNICAL.md)
+
+## Specifics
+
+- **mvn_cmd.rs** handles Maven (`mvn`), Maven Wrapper (`mvnw`), and Maven Daemon (`mvnd`) commands
+- `rtk mvn`: auto-detects `mvnw` wrapper in project root; falls back to system `mvn`
+- `rtk mvnd`: always invokes the Maven Daemon (`mvnd`) — the wrapper is bypassed because `mvnd` is a separate long-lived JVM daemon; metrics are tracked as `mvnd <goal>` in `rtk gain` so mvn/mvnd savings stay separate
+- `mvn test` uses a state-machine parser (Preamble → Testing → Summary → Done) for 97-99%+ savings on real-world output
+- `mvn verify` shares the same state-machine filter as `test`; surefire + failsafe `T E S T S` blocks accumulate into one combined summary. This is the canonical goal that produces `target/failsafe-reports/` (integration tests), so XML enrichment surfaces both unit- and integration-test failures
+- `mvn compile` uses line filtering to strip `[INFO]` noise, download progress, JVM/native-access warnings, and plugin chatter (jOOQ codegen, Liquibase, npm/React builds, typescript-generator). Also routes `process-classes` and `test-compile` through the same filter (same noise profile)
+- `mvn checkstyle:check` (aliased as `checkstyle`) compacts violation lines to `path:line:col [Rule] message`, strips mvn startup noise and Help-link boilerplate, keeps `N Checkstyle violations` summary and BUILD SUCCESS/FAILURE
+- `mvn dependency:tree` strips "omitted for duplicate" lines, "version managed from" annotations, and collapses deep transitive branches
+- `mvn clean` collapses to one line `mvn clean: deleted <path> (time)`; multi-module builds report `deleted N targets` (goal chains like `mvn clean verify` are handled by the multi-goal filter below)
+- **Single-goal routing** (`dispatch` → `route_goal`): `test`, `verify`, `clean`, compile-likes, `checkstyle`/`checkstyle:check`, and `dependency:tree` each route to their dedicated filter; any other goal (e.g. `package`, `install`, `deploy`, `spring-boot:run`) streams via unfiltered `cmd.status()` passthrough — safe for long-running goals
+- **Multi-goal chains** (≥2 goals, e.g. `mvn clean test-compile checkstyle:check`, `mvn clean verify`, `mvn clean install`): a signal-aware filter splits output by Maven plugin boundary markers, runs each segment group through its matching single-goal filter, and always preserves the BUILD signal — all `[ERROR]` lines, `BUILD SUCCESS`/`BUILD FAILURE` + `Total time`, per-module test counts, checkstyle violation counts, and (on failure) the Reactor Summary failing module. `-q`/`--quiet` is automatically stripped in multi-goal mode so rtk receives full output and does the compression itself. When the chain includes `test`/`verify`/`install`, surefire/failsafe XML enrichment applies exactly as for single-goal `verify`.
+- **0 goals** (e.g. `mvn -version`, `mvn --help`): unfiltered passthrough
+- **gradlew_cmd.rs** handles Gradle Wrapper (`gradlew`) command output
+
+## Output enrichment from Surefire/Failsafe XML reports
+
+When `mvn test` (or verify/integration-test) reports failures, rtk reads
+`target/surefire-reports/TEST-*.xml` and `target/failsafe-reports/*.xml`
+**after** the build finishes and appends a structured Failures section
+with:
+
+- Full stack trace per failure, with framework frames collapsed and the
+  root-cause segment preserved (up to 50 lines per trace).
+- Captured stdout + stderr from failing tests only, capped at 2000 chars
+  per test and 10000 chars total.
+- File counters in the footer: `(reports: N surefire, M failsafe, K stale files skipped)`.
+
+### Application-package detection
+
+rtk classifies stack frames as *application* vs *framework* by comparing
+frame class names against the Java `groupId` from `pom.xml`:
+
+1. `<project>/<groupId>` from the pom.xml in the current working directory.
+2. Fallback: `<project>/<parent>/<groupId>`.
+3. Otherwise: no filtering — full stack traces are preserved.
+
+### Time-gated report reads
+
+Stale XML reports from previous runs are skipped: only files with
+`mtime >= started_at` (captured just before `mvn` executes) are parsed.
+
+### Red-flag heuristic for "0 tests"
+
+If the summary says `no tests run` but surefire reports are empty or
+absent, rtk emits a diagnostic instead of the silent summary:
+
+```
+mvn test: 0 tests executed — surefire detected no tests. Check pom.xml (surefire plugin configuration) or run: rtk proxy mvn test
+```
+
+### Bypass
+
+For the rare cases where you need the full raw Maven output:
+
+```bash
+rtk proxy mvn test
+```