together-ai-inference

Serverless inference, fine-tuning, embeddings, image generation, and batch processing on 200+ open-source models via an OpenAI-compatible API. Use when you need fast, cost-effective access to open-source LLMs without managing infrastructure.

Quality

75%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./backend/cli/skills/cloud-compute/together-ai/SKILL.md

Quality

Content

71%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.

Highly actionable content with executable examples for every feature and clearly sequenced workflows, undermined by monolithic structure and notable padding — repeated client boilerplate, duplicated CLI sections, and reference tables inlined where progressive disclosure expects separate files.

Suggestions

Split reference material into bundle files (e.g., references/models.md for the model/pricing catalogs, references/cli.md for the CLI reference, references/fine-tuning.md for the full fine-tuning guide) and keep SKILL.md as a concise overview with clearly signaled one-level-deep links.

Remove the repeated 'from together import Together; client = Together()' boilerplate by stating the client setup once and referencing it, and deduplicate the CLI fine-tuning commands that appear in both the Fine-Tuning and CLI Reference sections.

Add an explicit validation step before submitting fine-tuning and batch jobs (e.g., verify each JSONL line parses and has required fields before upload) rather than only covering format errors in the troubleshooting table.

Dimension	Reasoning	Score
Conciseness	Dense reference material with little concept over-explanation, but noticeably padded: 'from together import Together; client = Together()' boilerplate is repeated in roughly fifteen snippets, install instructions appear twice, and CLI fine-tuning commands are duplicated verbatim across two sections, with DeepSeek-R1 and Qwen3-Coder priced in two tables each.	3 / 5
Actionability	Fully executable, copy-paste-ready code covering the common cases: chat, streaming, function calling, JSON mode, vision, fine-tuning (SDK and CLI), embeddings, image generation, batch submission, OpenAI/LangChain compatibility, plus a concrete troubleshooting table with specific fixes.	5 / 5
Workflow Clarity	Multi-step workflows are clearly sequenced (data format → upload → create job → monitor → download for fine-tuning; upload → create → check status → download for batch) with an explicit status checkpoint ('if status.status == "COMPLETED"') and a Common Issues table for error recovery, though pre-upload data-format validation is only addressed retroactively in troubleshooting.	4 / 5
Progressive Disclosure	Well-sectioned with clear headers and useful external doc links, but it is a 725-line monolith with no bundle files: the model catalogs, pricing tables, embedding/image model lists, and full CLI reference are inlined content that clearly belongs in separate reference files.	3 / 5
	Total	15 / 20 Passed

Description

80%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.

A strong description with explicit what/when structure and a comprehensive capability list, held back by second-person voice in the trigger clause and the absence of the platform name, which creates real conflict risk against other OpenAI-compatible open-source inference providers.

Suggestions

Name the platform explicitly in the description (e.g., 'Together AI: serverless inference...') so it is distinguishable from other OpenAI-compatible open-source model providers like Fireworks or Groq.

Rewrite the trigger clause in third person to avoid the second-person penalty, e.g., 'Use when the user needs fast, cost-effective access to open-source LLMs...'.

Add one or two natural synonyms users would say, such as 'chat completions' or model family names (Llama, DeepSeek, Qwen), to improve trigger term coverage.

Dimension	Reasoning	Score
Specificity	Enumerates five concrete capabilities ('Serverless inference, fine-tuning, embeddings, image generation, and batch processing on 200+ open-source models via an OpenAI-compatible API'), matching the comprehensive anchor 5, but the trigger clause 'Use when you need...' uses second person voice, which the guidelines penalize by one point.	4 / 5
Completeness	Explicitly answers both what ('Serverless inference, fine-tuning, embeddings, image generation, and batch processing...') and when ('Use when you need fast, cost-effective access to open-source LLMs without managing infrastructure'), with concrete trigger phrases.	5 / 5
Trigger Term Quality	Good natural keyword coverage ('inference', 'fine-tuning', 'embeddings', 'image generation', 'batch', 'open-source LLMs') that users would plausibly say, but common variations like 'chat completions', 'hosted models', or model family names (Llama, DeepSeek) are missing.	4 / 5
Distinctiveness Conflict Risk	The description never names 'Together AI', so its triggers ('serverless inference on open-source models via an OpenAI-compatible API') overlap heavily with similar provider skills such as Fireworks, Groq, or DeepInfra; it is somewhat specific but not clearly distinguishable.	3 / 5
	Total	16 / 20 Passed

Validation

87%

Warnings & errors only

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

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (736 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/cloud-compute/together-ai/SKILL.md
Commit: 4082a2e

Reviewed: 1 day ago

Table of Contents

Discovery Implementation Validation

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.