Gpt

Generate GPT API payloads for chat, embeddings, and fine-tuning. Estimate costs and batch requests.

Security Scan

The skill's code, instructions, and requirements are consistent with its goal of generating GPT API payloads, estimating costs, batching, and converting data — it doesn't request credentials or perform unexpected network or privileged actions.

Purpose & Capability

Name/description (generate chat/embedding/fine-tune payloads, estimate costs, batch/convert) match the included Bash script and SKILL.md. No unrelated credentials, binaries, or config paths are required.

Instruction Scope

SKILL.md only instructs running the provided script with files/inputs to produce JSON/JSONL outputs and cost estimates. The script reads user-specified files (CSV, text) and writes outputs, which is expected for this purpose; it does not instruct reading system config or credentials.

Install Mechanism

No install spec — instruction-only with a single included script. Nothing is downloaded or installed by the skill.

Credentials

The skill declares no required environment variables or credentials. The script does not reference environment secrets in the visible content.

Persistence & Privilege

always is false and the skill does not request persistent system presence or modify other skills. It runs as a local script invoked by the agent.

Assessment

This skill appears coherent and safe for its stated purpose. Before using it: (1) review the full scripts/script.sh file locally (the provided file is the primary behavior surface) to confirm there are no network calls in the truncated portion; (2) only pass non-sensitive files to the tool (it reads any input files you give it and writes outputs); (3) if you will use it to prepare fine-tuning data, scrub PII and verify JSONL output; (4) run the script locally in a sandbox or review it in a text editor if you have any doubt. If you need certainty that no network exfiltration exists, request the remainder of the script to be shown and inspected.

GPT API calling assistant. Build chat completion requests, generate embedding payloads, prepare fine-tuning data, estimate costs, batch requests, and convert between formats.

Commands

chat

Build a chat completion request JSON payload from messages.

bash scripts/script.sh chat --system "You are a translator" --user "Translate: hello"
bash scripts/script.sh chat --system "Helper" --user "Hi" --model gpt-4o --temperature 0.7

embed

Generate an embedding API request payload.

bash scripts/script.sh embed --input "Text to embed"
bash scripts/script.sh embed --file texts.txt --model text-embedding-3-small

finetune

Prepare and validate fine-tuning JSONL data from source files.

bash scripts/script.sh finetune --input pairs.csv --output training.jsonl
bash scripts/script.sh finetune --validate training.jsonl

cost

Estimate API costs based on token count and model pricing.

bash scripts/script.sh cost --tokens 5000 --model gpt-4o
bash scripts/script.sh cost --file input.txt --model gpt-4-turbo

batch

Generate multiple API request payloads from a list of inputs.

bash scripts/script.sh batch --file prompts.txt --model gpt-4o --output batch_requests.jsonl

convert

Convert between JSONL and CSV data formats.

bash scripts/script.sh convert --input data.csv --to jsonl --output data.jsonl
bash scripts/script.sh convert --input data.jsonl --to csv --output data.csv

Output

All commands output JSON request bodies or JSONL to stdout by default. Use --output to write to a file. Cost estimates print a breakdown table with per-model pricing.

Requirements

• bash 4+

Feedback

Report issues or suggestions: https://bytesagain.com/feedback/

---

Powered by BytesAgain | bytesagain.com