CtrlK
BlogDocsLog inGet started
Tessl Logo

deepgram-java-conversational-stt

Use when writing or reviewing Java code in this repo that calls Deepgram Conversational STT v2 / Flux over `/v2/listen`. Covers `client.listen().v2().v2WebSocket()`, `V2ConnectOptions`, `onTurnInfo`, and turn-aware close handling. Use `deepgram-java-speech-to-text` for standard v1 transcription and `deepgram-java-voice-agent` for fully interactive assistants. Triggers include "flux", "conversational stt", "listen v2", "turn detection", "end of turn", and "eot".

75

Quality

93%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

86%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-structured, SDK-specific reference skill: lean body, executable quick start, and clearly signaled one-level-deep references. The main weaknesses are a commented-out media-sending step and the absence of an explicit error-recovery loop in the workflow.

Suggestions

Wire `sendMedia(...)` with a concrete audio source (e.g., reading bytes from an `InputStream` in a loop) instead of leaving it commented out, so the quick start is end-to-end runnable rather than just handler/close wiring.

Add an explicit error-recovery checkpoint: register `onErrorMessage(...)` in the quick start and show the reaction (e.g., log the error and reconnect), giving the workflow a feedback loop.

Wrap `connect(...).get(...)` in a try/catch for `ExecutionException`/`TimeoutException` with a reconnect-retry step so the connect-before-send checkpoint has an explicit failure path.

DimensionReasoningScore

Conciseness

The body is lean and SDK-specific — it never explains general concepts (what WebSocket/STT is) and every section earns its place; the gotchas are all non-obvious, library-specific facts (model is a String not enum, sendCloseStream vs v1 finalize, no language_hint). It assumes Claude's competence throughout.

5 / 5

Actionability

The quick start gives a concrete, mostly copy-paste Java example with real imports, connect options, handlers, and close flow; the gap is that `sendMedia(...)` is only shown commented-out and audio streaming is hand-waved as "stream binary audio manually", leaving the end-to-end path incomplete.

4 / 5

Workflow Clarity

The quick start sequences build → wire handlers → connect (with `.get(10, TimeUnit.SECONDS)` wait) → send media → close, and gotcha #7 makes the connect-before-send checkpoint explicit; however `onErrorMessage(...)` is listed in the API surface but never wired, so there is no error-recovery feedback loop.

4 / 5

Progressive Disclosure

Clear sectioned overview with well-signaled one-level-deep references — the numbered "API reference (layered)" points to in-repo source (`src/main/java/...`, `examples/listen/LiveStreamingV2.java`) and external docs (AsyncAPI, Context7, product docs) with no nested/2+ level reference chains, and content is appropriately kept inline for a ~100-line skill.

5 / 5

Total

18

/

20

Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

An excellent description: it states a precise domain and API surface, gives an explicit "Use when" clause with concrete trigger phrases, and draws clear boundaries against two sibling skills. It avoids vague language and over-claims entirely.

DimensionReasoningScore

Specificity

Names the concrete domain ("writing or reviewing Java code ... that calls Deepgram Conversational STT v2 / Flux over `/v2/listen`") and lists multiple specific API symbols — `client.listen().v2().v2WebSocket()`, `V2ConnectOptions`, `onTurnInfo`, "turn-aware close handling" — giving comprehensive coverage of the API surface rather than vague language.

5 / 5

Completeness

Explicitly answers both "what" (covers specific v2/Flux API surfaces and turn-aware close handling) and "when" ("Use when writing or reviewing Java code ..." plus a concrete "Triggers include ..." list), matching the clearly-answers-both-with-trigger-phrases anchor.

5 / 5

Trigger Term Quality

Explicitly enumerates natural trigger phrases — "flux", "conversational stt", "listen v2", "turn detection", "end of turn", "eot" — including the synonym pair ("end of turn" / "eot") and the product/version name, matching the comprehensive-synonyms anchor.

5 / 5

Distinctiveness Conflict Risk

Occupies a clear niche (Java + Conversational STT v2/Flux over `/v2/listen`) and explicitly redirects to siblings — `deepgram-java-speech-to-text` for v1 and `deepgram-java-voice-agent` for assistants — minimizing trigger overlap with adjacent skills.

5 / 5

Total

20

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
deepgram/deepgram-java-sdk
Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.