prompt-engineering

prompt-engineering is a skill for Claude Code, Codex from int2t05/engineering-skills. It costs 73 tokens per session (1,612 once invoked), scanned A, original, MIT.

A guide for designing prompts and evaluations for features that use large language models. It treats the prompt, model choice, safety rules, and tests that measure quality as one engineering task.

In plain words
What is it for?
Use it when building chat, summarisation, extraction, classification, or generation features. It helps define success criteria, choose models, add safeguards, and create an evaluation harness.
Why use it?
It replaces guesswork with explicit goals and repeatable checks. Evaluations show whether a prompt or model actually performs well instead of relying on a few impressions.

Skill for Claude CodeCodex

Written for Claude Code and Codex: ${CLAUDE_PLUGIN_ROOT} variable, but also agents/openai.yaml present.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the engineering-skills plugin — 47 skills, 1 hook shipped together

Good fit Use it when building chat, summarisation, extraction, classification, or generation features.

Compare 6 skills from other repositories ↓
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add int2t05/engineering-skills
Claude Code
/plugin install engineering-skills

Made for: Claude Code, Codex.

Or install engineering-skills, the plugin that ships this one along with the rest of its 47 skills, 1 hook.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for prompt-engineering

README.md
[![agentmods](https://agentmods.dev/badge/skills/int2t05/engineering-skills/prompt-engineering.svg)](https://agentmods.dev/skills/int2t05/engineering-skills/prompt-engineering)
Your own site
<a href="https://agentmods.dev/skills/int2t05/engineering-skills/prompt-engineering"><img src="https://agentmods.dev/badge/skills/int2t05/engineering-skills/prompt-engineering.svg" alt="Measured on agentmods" height="20"></a>
Per session 73 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,612 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00073 $0.01612
Opus 5 $0.00036 $0.00806
Sonnet 5 $0.00015 $0.00322
Haiku 4.5 $0.00007 $0.00161

Measured 6d ago against content hash 05e0d419eeb5, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

prompt-engineering scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/03-design/prompt-engineering/SKILL.md · 133 lines

How it starts

The opening of the file, as written. The whole thing — 133 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Prompt Engineering

Design prompts and evals as a first-class engineering deliverable. context-engineering assembles context for a coding agent; this skill designs the prompt-as-product surface — the prompts, model choices, guardrails, and eval harnesses behind LLM-powered features. A prompt without an eval is an opinion; an eval without a prompt is a benchmark. Ship neither blind.

When to use

  • Designing an LLM-powered feature (chat, summarization, extraction, classification, generation)
  • Building or refining a prompt for production use
  • Creating an eval harness to measure prompt/model quality
  • Selecting a model for a specific task against cost/latency/quality trade-offs
  • Triggers on "prompt engineering", "LLM feature", "eval harness", "prompt design", "提示词工程", "LLM 特性"

Not for: assembling context for a coding agent (use context-engineering); general research on a topic (use research); API contract design for non-LLM endpoints (use api-design).

Steps

1. Define the task and success criteria

State the task in one sentence, then define measurable success criteria — without these, prompt iteration is vibes-driven. Pull from the product spec:

  • Input space: what inputs will the prompt receive? (vary by length, language, edge case, adversarial)
  • Output contract: structured output (JSON schema), free text, or classification?
  • Quality bar: accuracy %, format adherence %, hallucination rate, latency target, cost per call
  • Failure modes to prevent: what must the model NEVER do? (leak PII, invent facts, refuse valid input)

Verify: success criteria are written as measurable thresholds, not "good responses."

2. Design the prompt architecture

Structure the prompt as components, not a wall of text — each component has a job:

  • System / role: who the model is, what it must and must not do (guardrails live here)
  • Task instruction: the operation, stated precisely with the output format
  • Context / retrieved data: only the facts the model needs (not the whole knowledge base — context bloat degrades accuracy and raises cost)
  • Few-shot examples: 2–5 input→output pairs covering the happy path and an edge case; place before the actual input, after the instruction
  • Output format: explicit schema or template; use structured output (JSON mode / function calling) when the downstream system parses the result

Read the full file on GitHub · 133 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 6d ago First seen · 133 lines · 73 tokens per session scan A 05e0d419eeb5

Subscribe to this mod's changes

prompt-engineering is a skill published in the GitHub repository int2t05/engineering-skills (3 stars, last pushed 5d ago), licensed MIT. It adds 73 tokens to every session and 1,612 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

pr-workflow

Crear pull requests completas con descripcion, labels y reviewers.

686f6c61/alfred-dev · 14 tokens

review-prompt

Review LLM prompts against the prompt-engineering skill's principles — leading with where each line came from — and report the findings without modifying files. Use when reviewing prompt quality, auditing a prompt, evaluating a system prompt, or checking whether prompt issues are high-confidence and fixable.

doodledood/manifest-dev · 60 tokens

goal-test

A local experiment for testing a goal command that keeps an AI coding session working until a stated condition is judged complete. It uses a separate language model to evaluate the conversation after each assistant turn.

restarter/lets-workflow · 137 tokens

llm

AI features you ship to users: structured output, tool schemas, prompt injection, evals. Use when "the model returns bad JSON", "it hallucinates", "stop it calling the wrong tool", "add evals", or an LLM feature can trigger refunds, emails or writes. Covers schema-constrained output, idempotent tool calls…

rainmanjam/poka-yoke · 89 tokens

fec-drawio-studio

An editable workflow for making technical diagrams in draw.io (also called diagrams.net), with the original .drawio file saved alongside exported images or documents. It covers architecture, database, UML, sequence, flow, machine-learning, and code-structure diagrams.

bovinphang/frontend-craft · 154 tokens

pm

Project manager for GitHub issues: brainstorm design approaches, create structured issues optimized for LLM agent teams, triage and recommend what to work on next, audit and clean up stale issues, or deep-validate a single issue against the codebase. Triggers: create issue, plan work, new task, project manager, write…

rube-de/cc-skills · 98 tokens