deepgram-dotnet-conversational-stt
Use when evaluating or extending conversational / Flux-style streaming STT support in this C# SDK. The repo has a latest WebSocket listen client under `Deepgram.Models.Listen.v2.WebSocket`, but it does not currently expose Flux-specific request params or turn-aware response types like `TurnInfo`. Use this skill to document that gap honestly and to guide work on the closest supported surface.
80
70%
Does it follow best practices?
Impact
96%
1.06xAverage score across 3 eval scenarios
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./.agents/skills/deepgram-dotnet-conversational-stt/SKILL.mdQuality
Discovery
75%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
The description has strong completeness with an explicit 'Use when' clause and is highly distinctive due to its narrow domain focus on the Deepgram C# SDK's Flux streaming STT support. However, the actual capabilities described are somewhat vague ('document that gap' and 'guide work') rather than listing concrete actions, and the trigger terms lean heavily technical without covering common user phrasings for speech-to-text tasks.
Suggestions
List more concrete actions the skill performs, e.g., 'Identifies missing Flux request parameters, maps existing WebSocket response types, generates wrapper code for TurnInfo support'.
Add natural-language trigger terms users might say, such as 'speech-to-text', 'real-time transcription', 'voice streaming', or 'Deepgram .NET SDK'.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | It names the domain (C# SDK, WebSocket streaming STT) and mentions some specific elements like 'Flux-style streaming STT', 'TurnInfo', and 'Flux-specific request params', but the actual actions are vague—'document that gap honestly' and 'guide work on the closest supported surface' are not concrete, enumerable actions. | 2 / 3 |
Completeness | The description explicitly answers both 'what' (evaluating/extending Flux-style streaming STT support, documenting gaps in the SDK) and 'when' ('Use when evaluating or extending conversational / Flux-style streaming STT support in this C# SDK'), with a clear 'Use when...' clause at the beginning. | 3 / 3 |
Trigger Term Quality | Includes some relevant keywords like 'STT', 'Flux', 'WebSocket', 'C# SDK', 'Deepgram', 'streaming', and 'TurnInfo', but these are fairly technical and niche. Common user terms like 'speech-to-text', 'real-time transcription', or 'voice recognition' are missing. | 2 / 3 |
Distinctiveness Conflict Risk | Highly specific niche—Deepgram C# SDK, Flux-style streaming STT, WebSocket listen client with specific model paths. Very unlikely to conflict with other skills due to the narrow domain focus. | 3 / 3 |
Total | 10 / 12 Passed |
Implementation
64%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill does a good job honestly documenting a gap in the .NET SDK's Flux/conversational STT support, with a strong executable code example for the closest available path. Its main weaknesses are the lack of a clear step-by-step workflow for extending the SDK to add Flux support (the most likely reason someone would use this skill) and some token inefficiency in sections that explain things Claude already knows or provide marketing-style cross-references.
Suggestions
Add an explicit, sequenced workflow with validation checkpoints for the task of extending the SDK to add Flux support (e.g., 1. Add TurnInfo model → 2. Extend LiveSchema with Flux params → 3. Write unit tests → 4. Validate against AsyncAPI spec → 5. Add example).
Trim the 'When to use this product' section and 'Central product skills' section to reduce token usage — Claude can route between skills without detailed routing instructions, and the npx install block is not actionable for the SDK gap documentation task.
Move the detailed API reference links and example file listings into a separate REFERENCES.md to improve progressive disclosure and keep the main skill focused on the gap documentation and actionable code.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is mostly efficient and documents a genuine gap clearly, but includes some unnecessary sections like the 'When to use this product' routing guidance and the 'Central product skills' marketing-style section that add tokens without much actionable value. The authentication section explains basic package installation which Claude already knows. | 2 / 3 |
Actionability | Provides fully executable, copy-paste-ready C# code for the closest supported code path, lists concrete file paths for source-of-truth references, enumerates specific available parameters on LiveSchema, and gives explicit guidance on what does and does not exist. The code example is complete and runnable. | 3 / 3 |
Workflow Clarity | The skill clearly documents what exists vs. what doesn't and provides a working code path, but there's no explicit workflow for the most likely task (extending the SDK to add Flux support). Gotcha #4 mentions 'start by adding models, request params, examples, and tests' but doesn't sequence these steps or provide validation checkpoints for what would be a multi-step, potentially destructive modification process. | 2 / 3 |
Progressive Disclosure | The skill has reasonable section structure and references external files (source paths, example files, external docs), but all content is inline in a single file with no bundle files to offload detail. The API reference section with 5 numbered categories and multiple URLs could be better organized in a separate reference file. The layered references are well-signaled but the skill is somewhat long for what it conveys. | 2 / 3 |
Total | 9 / 12 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.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- deepgram/deepgram-dotnet-sdk
- Commit
d0d6fee
- 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.