skill(apm-integrations): match the sibling module's test DSL (no Java tests in a Groovy family)

[jordan-wong]

🤖 Generated with APM Instrumentation Toolkit

[Reference / skill-fix PR — draft.] From the toolkit eval-research thread; a sharpening of the apm-integrations agent skill, not an instrumentation change.

What

Fixes a self-contradiction in references/tests.md:

• Line 7 mandates "Write Groovy/Spock tests ... Full Java instrumentation test support is not yet available."
• Lines ~136–175 illustrate the no-banner-comments rule with a fully Java AbstractInstrumentationTest example, and nothing states which wins.

Adds an unambiguous rule — match the sibling module's test DSL; a Groovy family stays Groovy — and fences the Java examples as style-only illustrations for modules already on the Java/JUnit DSL (not a license to migrate a Groovy family to Java).

Why (evidence)

Surfaced by the 2026-08-07 cache-category gap sweep on jedis-3.0 (reference PR #12080). jedis-3.0 is otherwise clean, but the regen wrote its tests in Java/JUnit (src/test/java, AbstractInstrumentationTest) while master and its siblings (jedis-1.4, jedis-4.0) use Groovy/Spock (src/test/groovy). The generation transcript shows the agent read the Groovy prose and a sibling cache library's Java test (lettuce-5.0) and followed the Java exemplar.

That Java choice is the direct trigger of the spurious _dd.svc_src CI failure a prior report mis-attributed to pure master-side infra: the Java DSL's TagsMatcher.defaultTags() (which carries a // TODO Keep porting default tag logic gap) enforces exact default tags the Groovy DSL tolerates. Fixing the skill (so the family stays Groovy) is higher-leverage than patching TagsMatcher.

Domain-agnostic: the contradiction is in the shared skill, so any Java instrumentation generation faces the same fork. Surfaced via cache but not cache-specific.

Scope

One file, references/tests.md, +3/−1. No instrumentation code changes.

🤖 Generated with APM Instrumentation Toolkit

[dd-octo-sts]

🟢 Java Benchmark SLOs — All performance SLOs passed

Suite	Status
Startup	🟢 pass

SLO thresholds are defined here based on automatically generated metrics. A warning is raised when results are within 5% of the threshold.

PR vs. master results

Scenario	Candidate	master	Δ (95% CI of mean)
startup:insecure-bank:iast:Agent	14.01 s	13.97 s	[-0.6%; +1.3%] (no difference)
startup:insecure-bank:tracing:Agent	13.02 s	13.00 s	[-0.8%; +1.2%] (no difference)
startup:petclinic:appsec:Agent	17.47 s	17.32 s	[-0.3%; +2.0%] (no difference)
startup:petclinic:iast:Agent	16.74 s	17.46 s	[-8.3%; +0.1%] (no difference)
startup:petclinic:profiling:Agent	17.42 s	17.39 s	[-1.1%; +1.5%] (no difference)
startup:petclinic:sca:Agent	17.33 s	17.32 s	[-0.8%; +0.9%] (no difference)
startup:petclinic:tracing:Agent	16.13 s	16.65 s	[-7.3%; +0.9%] (no difference)

Commit: 7c00fb4e · CI Pipeline · Benchmarking Platform UI

---

Load and DaCapo benchmarks can be triggered manually in the GitLab pipeline. Results will appear in the Benchmarking Platform UI after completion.