Java library (Maven Central-ready)

A Java library for other JVM code to depend on, built with Maven and make. A JPMS module exporting one package, JSpecify nullness checked by NullAway, Javadoc with doclint and examples that run as tests, generated-input property tests, a consumer project built against the installed jar, a japicmp API check and the sources and Javadoc jars Maven Central requires.

What done means

Your agent follows the brief, then runs stacks verify without weakening or skipping checks. What happens next

Declared steps: install, fmt-check, lint, compile, test, docs, build, consumer, api-check, readme-filled. View commands

Use this stack

Paste the prompt into the coding agent working on your project.

Full projectQuick local experiment

Uses the stack’s default fragments. Review the brief with your agent.Uses stated defaults and omits deploy, release, and app fragments unless explicitly requested.

View prompt

Use the java-library stack for this project. 1. Run npx @steamfield/stacks show java-library --json and check that it fits the project. If it does not, explain why before changing files. 2. Run npx @steamfield/stacks new java-library and follow the brief it prints. 3. From the generated project directory, run npx @steamfield/stacks verify. Fix failures without weakening or skipping verification steps. 4. Summarize what changed and the verification result. Quote published stats only when they are available, including their interval and sample size. The stacks docs, written for agents, are at https://stacks.steamfield.io/llms.txt.

Use the java-library stack for a quick local experiment. 1. Run npx @steamfield/stacks show java-library --json and check that it fits the project. If it does not, explain why before changing files. 2. Run npx @steamfield/stacks new java-library --sandbox and follow the brief it prints. 3. From the generated project directory, run npx @steamfield/stacks verify. Fix failures without weakening or skipping verification steps. 4. Summarize what changed and the verification result. Quote published stats only when they are available, including their interval and sample size. The stacks docs, written for agents, are at https://stacks.steamfield.io/llms.txt.

Or use the terminal

npx @steamfield/stacks new java-library

npx @steamfield/stacks new java-library --sandbox

Verify steps
Step	Phase	Layer	Command
install	setup	install	`make install`
fmt-check	check	lint	`make fmt-check`
lint	check	lint	`make lint`
compile	check	typecheck	`make compile`
test	check	test	`make coverage`
docs	check	docs	`make docs`
build	check	build	`make build`
consumer	check	test	`make test-consumer`
api-check	check	check	`make api-check`
readme-filled	check	check	`test -s README.md && ! grep -n 'stacks:fill' README.md`

Usage and outcomes

Usage over time

Showing 180 days of history.

History

The 28-day green rate at each weekly observation. Gaps mean the sample was too small to publish.

Shaded band: 95% Wilson interval. Dashed rules: content changes. Missing observations remain gaps.

Stacks projects · 28 days through 4 Nov 2026. Counts rounded to 5.

Data table

Through	Green rate	Projects	Content change
5 Jul 2026	hidden: too few projects	–	–
12 Jul 2026	76% (67–83%, n=105)	145	–
19 Jul 2026	79% (73–84%, n=200)	290	–
26 Jul 2026	77% (72–81%, n=355)	535	–
2 Aug 2026	77% (73–81%, n=535)	800	–
9 Aug 2026	78% (74–81%, n=600)	915	–
16 Aug 2026	76% (73–79%, n=705)	1,060	–
23 Aug 2026	76% (73–79%, n=730)	1,090	–
30 Aug 2026	75% (71–77%, n=795)	1,140	–
6 Sept 2026	74% (71–77%, n=850)	1,215	–
13 Sept 2026	74% (71–77%, n=895)	1,305	–
20 Sept 2026	75% (72–77%, n=940)	1,375	–
27 Sept 2026	74% (71–77%, n=975)	1,455	–
4 Oct 2026	74% (71–76%, n=1,030)	1,550	–
11 Oct 2026	73% (70–76%, n=1,095)	1,620	–
18 Oct 2026	74% (71–76%, n=1,185)	1,730	–
25 Oct 2026	74% (71–76%, n=1,250)	1,840	–
1 Nov 2026	73% (71–76%, n=1,310)	1,930	–

Weekly projects by preset

Non-overlapping seven-day project counts, grouped by the preset used to start the project.

default
sandbox

Stacks projects per UTC week through 4 Nov 2026. Published preset counts only; hidden cells are not zero.

Data table

Week ending	Preset	Projects
5 Jul 2026	default	–
12 Jul 2026	default	90
19 Jul 2026	default	110
26 Jul 2026	default	155
2 Aug 2026	default	170
9 Aug 2026	default	160
16 Aug 2026	default	205
23 Aug 2026	default	170
30 Aug 2026	default	200
6 Sept 2026	default	225
13 Sept 2026	default	240
20 Sept 2026	default	215
27 Sept 2026	default	270
4 Oct 2026	default	260
11 Oct 2026	default	285
18 Oct 2026	default	300
25 Oct 2026	default	350
1 Nov 2026	default	335
5 Jul 2026	sandbox	–
12 Jul 2026	sandbox	55
19 Jul 2026	sandbox	40
26 Jul 2026	sandbox	85
2 Aug 2026	sandbox	95
9 Aug 2026	sandbox	100
16 Aug 2026	sandbox	85
23 Aug 2026	sandbox	105
30 Aug 2026	sandbox	120
6 Sept 2026	sandbox	110
13 Sept 2026	sandbox	135
20 Sept 2026	sandbox	125
27 Sept 2026	sandbox	135
4 Oct 2026	sandbox	165
11 Oct 2026	sandbox	160
18 Oct 2026	sandbox	155
25 Oct 2026	sandbox	165
1 Nov 2026	sandbox	180

By preset

Green verify outcomes by the preset chosen at scaffold time.

default

70%67–73%, n=800

sandbox

79%75–83%, n=415

Stacks projects · 28 days through 4 Nov 2026. Counts rounded to 5.

Data table

Name	Green rate	Projects
default	70% (67–73%, n=800)	1,170
sandbox	79% (75–83%, n=415)	620

By agent

Within this stack, listed alphabetically. These are project outcomes, not an agent ranking.

claude

73%70–76%, n=660

codex

70%65–75%, n=305

cursor

77%69–83%, n=125

none

73%57–84%, n=40

opencode

79%68–87%, n=70

otherhidden: too few projects

Stacks projects · 28 days through 4 Nov 2026. Counts rounded to 5.

Data table

Name	Green rate	Projects
claude	73% (70–76%, n=660)	970
codex	70% (65–75%, n=305)	455
cursor	77% (69–83%, n=125)	185
none	73% (57–84%, n=40)	55
opencode	79% (68–87%, n=70)	105
other	hidden: too few projects	25

Time to green

Published time summaries on a shared minutes scale. First passing test is a median; 75% reached green by the 75th percentile.

Verify runs to green: 1

Stacks projects · 28 days through 4 Nov 2026. Counts rounded to 5.

Data table

Measure	Time / runs
Median to green	22 minutes
75th percentile	31 minutes
First passing test	13 minutes
Verify runs to green	1

Where it breaks

A run can fail more than one step. Environment failures are shown separately.

test25%

lint15%

build12%

readme-filled9%

compile8%

fmt-check7%

docs7%

api-check6%

consumer5%

install5%

Environment blocked0%

Share of failed runs before first green. Environment blocked describes the machine, not the stack.

Data table

Step	Share
build	12%
test	25%
api-check	6%
fmt-check	7%
consumer	5%
readme-filled	9%
install	5%
compile	8%
lint	15%
docs	7%
Environment blocked	0%

What people change

Defaults people keep, and optional fragments they choose to attach.

Default, kept rate

agent-ops92%

git-hooks92%

ci-github92%

Optional, attach rate

rr18%

ce-plugin17%

Share of the window's 1,795 projects. Kept rate is 1 minus the without rate.

Data table

Fragment	Measure	Share
agent-ops	Kept	92%
git-hooks	Kept	92%
ci-github	Kept	92%
rr	Attached	18%
ce-plugin	Attached	17%

By composition 23 fingerprints

Each row is a distinct fragment composition, with its own sample and content history.

Composition	Green rate	Seen	Change in life
defaults + rr − agent-ops · 02c0b56b	hidden: too few projects	12 Jul 2026 – 1 Nov 2026	None
defaults · 033047f6	75% (72–78%, n=830)	12 Jul 2026 – 1 Nov 2026	None
defaults + ce-plugin − git-hooks · 0f187ed2	hidden: too few projects	26 Jul 2026 – 1 Nov 2026	None
defaults + rr − ci-github − git-hooks · 2f1c1c3f	hidden: too few projects	6 Sept 2026 – 1 Nov 2026	None
defaults + rr − agent-ops − ci-github · 4510d892	hidden: too few projects	30 Aug 2026 – 1 Nov 2026	None
defaults + ce-plugin − agent-ops − git-hooks · 4c11b86f	hidden: too few projects	6 Sept 2026 – 1 Nov 2026	None
defaults + rr − git-hooks · 5d749986	hidden: too few projects	12 Jul 2026 – 1 Nov 2026	None
defaults − git-hooks · 725da01f	hidden: too few projects	12 Jul 2026 – 1 Nov 2026	None
defaults + rr − agent-ops − git-hooks · 853b9544	hidden: too few projects	9 Aug 2026 – 1 Nov 2026	None
defaults + ce-plugin + rr − agent-ops · 8aa6081c	hidden: too few projects	16 Aug 2026 – 1 Nov 2026	None
defaults − ci-github · 9f782ee2	hidden: too few projects	12 Jul 2026 – 1 Nov 2026	None
defaults + ce-plugin + rr − git-hooks · a265ed7b	hidden: too few projects	26 Jul 2026 – 1 Nov 2026	None
defaults + ce-plugin · a96b36fa	69% (62–75%, n=180)	12 Jul 2026 – 1 Nov 2026	None
defaults + ce-plugin − agent-ops · bc7714d8	hidden: too few projects	19 Jul 2026 – 1 Nov 2026	None
defaults − ci-github − git-hooks · c31b2dc6	hidden: too few projects	26 Jul 2026 – 1 Nov 2026	None
defaults + rr · c84c759b	71% (64–77%, n=175)	12 Jul 2026 – 1 Nov 2026	None
defaults − agent-ops − ci-github · ce2b648d	hidden: too few projects	12 Jul 2026 – 1 Nov 2026	None
defaults − agent-ops · d3ccacc2	hidden: too few projects	12 Jul 2026 – 1 Nov 2026	None
defaults + ce-plugin − ci-github · dbeafd66	hidden: too few projects	12 Jul 2026 – 1 Nov 2026	None
defaults + ce-plugin + rr · f0cc1281	68% (51–81%, n=35)	12 Jul 2026 – 1 Nov 2026	None
defaults + ce-plugin + rr − ci-github · f5df2a36	hidden: too few projects	2 Aug 2026 – 1 Nov 2026	None
defaults − agent-ops − git-hooks · f9e09177	hidden: too few projects	19 Jul 2026 – 1 Nov 2026	None
defaults + rr − ci-github · ffe208a5	hidden: too few projects	12 Jul 2026 – 1 Nov 2026	None

Computed 5 Nov 2026, through 4 Nov 2026. Numbers describe stacks projects, not the ecosystem.