✦ Prompteiro — AI Prompt Engineering, Inside VS Code

Write better prompts, get better AI output. Prompteiro supercharges every prompt you write directly inside Visual Studio Code — using private local AI models via Ollama, or your own remote API keys for OpenAI, Anthropic, Gemini, and Azure OpenAI.

Why Prompteiro?

Raw prompts produce mediocre results. Prompt engineering takes time. Prompteiro bridges both — it intelligently enhances your prompts using structured components (role, task, context, format, tone) so your AI tools understand exactly what you want. All of this happens in-editor with zero context switching.

🔒 100% Private in Local Mode — your prompts never leave your machine
⚡ Instant, in-editor — select text, click Enhance, done
🧠 Context-aware — automatically reads your file language, imports, surrounding code, and recent git commits to craft better prompts
🖥 Local or Remote — choose between fully offline Ollama models or your own API keys
🛡 Secure by design — API keys stored in OS keychain, never in settings files

Features

Select any text in your editor and a floating action toolbar appears inline — no menus, no command palette, no friction.

Action	Description
Enhance	Restructures your prompt with role, task, context, output format, tone, and length components for maximum LLM comprehension.
Clarify	Sharpens vague or ambiguous instructions — removes implicit assumptions and fills in missing details.
Add Context (coming soon)	Automatically injects surrounding code symbols, imports, and project context into your prompt.
Format Output (Pro)	Structures the output format expectation (e.g. JSON schema, markdown, table).
Save as Template (Pro)	Saves the enhanced prompt as a reusable template for future use.

✦ Side-by-Side Review Panel

When an enhancement completes, a split panel opens showing:

Original prompt (left)
Enhanced prompt (right)
Confidence score (0–100%) from the model
Component breakdown — what was changed and why
Token-by-token streaming — watch the result build in real time
Copy, Accept (replaces selection), or Dismiss controls

✦ Guided Local Setup (Ollama)

Prompteiro includes a full onboarding wizard:

Sign in with GitHub, Google, or Microsoft
Detects whether Ollama is installed — installs it for you if not (with a live download progress bar)
Recommends the optimal model for your machine's RAM
Downloads the selected model with a real-time progress bar
Launches and health-checks Ollama automatically

Supported local models:

Model	Size	Best For
`qwen2.5-coder:1.5b`	~1.0 GB	Laptops, fast iteration
`llama3:8b`	~4.7 GB	Balanced quality + speed
`llama3:70b`	~40 GB+	Maximum intelligence (high-RAM systems)

✦ Remote API Key Support

Connect your own API keys for cloud-based providers:

Provider	Models Supported
OpenAI	GPT-4o, GPT-4 Turbo, GPT-3.5-Turbo
Anthropic	Claude 3.5 Sonnet, Claude 3 Opus/Haiku
Google Gemini	Gemini 1.5 Pro, Gemini 1.5 Flash
Azure OpenAI	Your custom Azure-deployed models

Keys are stored securely in your OS keychain via VS Code's native SecretStorage API and are never transmitted to Prompteiro servers.

✦ Smart Context Extraction

When enhancing, Prompteiro automatically (and optionally) enriches your prompt with:

Active file language & framework
Surrounding code (up to 50 lines before/after selection)
Document symbols (functions, classes, interfaces in scope)
Import statements (top 5 detected)
Project type (Next.js, Rust/Cargo, Python, etc.)
Recent git commit history (last 5 commits)

Context extraction is bounded by a 500ms hard deadline to guarantee zero editor lag.

✦ Status Bar Integration

The status bar shows your active engine at all times:

🖥 Local: llama3:8b — Ollama running, model loaded
☁ OpenAI: gpt-4o — Remote mode active
⚠️ Ollama offline — Click to reconnect

✦ Entitlement System

Plan	Local Enhancements / Day	Remote Mode
Spark (Free)	15 per day	With your own API key
Ignite (Pro)	Unlimited	Included

Requirements

Visual Studio Code v1.90.0 or later
For Local Mode: Ollama installed on your machine (Prompteiro can install it automatically)
For Remote Mode: A valid API key from OpenAI, Anthropic, Google, or Azure OpenAI

Installation & Setup

Method 1: Extension Marketplace (Recommended)

Search for Prompteiro in the VS Code Extensions panel (Ctrl+Shift+X) and click Install.

Method 2: VSIX Install

code --install-extension prompteiro-0.1.6.vsix

First Launch

After installation:

The onboarding panel opens automatically.
Sign in with your GitHub, Google, or Microsoft account.
Choose Local AI (private, offline) or Remote API Key (cloud).
Follow the wizard — Prompteiro handles Ollama installation and model setup for you.

Commands

Open the Command Palette (Ctrl+Shift+P / Cmd+Shift+P) and type:

Command	Description
`Prompteiro: Enhance Prompt`	Enhance the currently selected text
`Prompteiro: Clarify Prompt`	Clarify the currently selected text
`Prompteiro: Setup & Login`	Reopen the onboarding and login panel

Error Codes Reference

If an operation fails, Prompteiro displays a structured error code so you can identify the root cause immediately:

Error Code	Meaning
`ERR_HTTP_STATUS_403`	Forbidden — check URL or authentication
`ERR_HTTP_STATUS_404`	Resource not found at the download URL
`ERR_HTTP_STATUS_5XX`	Server-side error at the remote host
`ERR_NET_ENOTFOUND`	No internet connection or DNS failure
`ERR_NET_ECONNRESET`	Connection was reset mid-transfer
`ERR_NET_ETIMEDOUT`	Connection timed out
`ERR_INSTALLER_EXIT_1`	Installer exited with non-zero code
`ERR_INSTALLER_LAUNCH_ENOENT`	Installer binary not found
`ERR_UNKNOWN`	Unexpected error — check extension logs

Security & Privacy

API Key Storage

All API keys and session tokens are stored exclusively in VS Code's native SecretStorage API, which uses your operating system's secure credential store (Windows Credential Manager, macOS Keychain, or libsecret on Linux). Keys are never written to:

VS Code settings.json
Extension global state
Any file on disk
Any Prompteiro server

Local Mode Privacy

In local mode (Ollama), 100% of prompt processing happens on your machine. No data is sent to any external server. The only network request is fetching the Ollama binary and model weights from official Ollama/HuggingFace servers.

Remote Mode Data Handling

When using remote mode with your own API key:

Your prompt text and optional code context are sent directly from your machine to the provider (OpenAI, Anthropic, etc.) using your key.
Prompteiro's servers are not in the request path.
Only what you explicitly select is sent — full workspace files are never scanned.

The first time code context extraction is used with a remote engine, Prompteiro shows an explicit consent dialog. You can withdraw consent at any time.

Workspace Trust

Prompteiro fully respects VS Code Workspace Trust:

In trusted workspaces: Full feature set including code context extraction.
In untrusted workspaces: Core enhancement works but file scanning, symbol extraction, and git log reading are disabled.

Telemetry

Telemetry is disabled by default. You can optionally enable anonymous usage statistics (no PII, no prompt content) via prompteiro.telemetry in VS Code settings.

Terms & Conditions

Last Updated: June 2026

By installing and using the Prompteiro VS Code Extension ("Extension"), you agree to the following terms:

1. Permitted Use

The Extension is provided for lawful personal and professional use. You may use it to enhance prompts for legitimate software development, writing, and research tasks. Misuse of the Extension to generate harmful, illegal, or deceptive content is strictly prohibited.

2. Local Mode

When operating in Local mode, all AI processing is performed exclusively on your local machine via Ollama. Prompteiro bears no responsibility for the behavior, accuracy, or output of locally-run AI models.

3. Remote Mode & Third-Party APIs

When you configure the Extension with a third-party API key (OpenAI, Anthropic, Google, Azure), your use of that API is governed solely by the terms of service of that provider. Prompteiro is not affiliated with and bears no responsibility for those providers' services, pricing, availability, or data handling.

4. Data Sent During Remote Enhancements

When remote mode is active and you trigger an enhancement, the following information may be sent directly to your configured AI provider:

The selected prompt text
Surrounding code context (up to 50 lines), if consent is granted
File language, project type, and import hints

This data is transmitted using your own API key directly to your provider. Prompteiro's servers do not receive or store this data.

5. Authentication

Sign-in is provided for quota management and account features. Authentication is handled via OAuth 2.0 through your chosen identity provider (GitHub, Google, or Microsoft). Prompteiro stores only a session token in VS Code SecretStorage, not your password.

6. Quotas & Entitlements

Free (Spark) plan users are entitled to 15 local AI enhancements per day. Quotas reset at midnight UTC. Circumventing quota enforcement (e.g., by modifying the extension or forging tokens) is a violation of these terms and may result in account suspension.

7. Intellectual Property

The Extension software, branding, and associated assets are the intellectual property of the Prompteiro project. You may not redistribute, resell, reverse-engineer, or create derivative works of the Extension without explicit written permission.

8. No Warranty

The Extension is provided "as is", without warranty of any kind, express or implied. Prompteiro does not guarantee uninterrupted service, accuracy of AI outputs, or fitness for any particular purpose. AI model outputs should be reviewed and validated before use in production systems.

9. Limitation of Liability

To the maximum extent permitted by law, Prompteiro and its contributors shall not be liable for any direct, indirect, incidental, special, or consequential damages arising from your use of the Extension or AI-generated content, even if advised of the possibility of such damages.

10. Changes to Terms

These terms may be updated from time to time. Continued use of the Extension after an update constitutes acceptance of the revised terms. The current version of these terms is always available in this README and at github.com/serpentxxx-cyrpt/Prompteiro.

11. Contact

For questions, bug reports, or security disclosures, please open an issue at: https://github.com/serpentxxx-cyrpt/Prompteiro/issues

Cloud Workspace & Sync (Ignite Tier)

If you are on the Ignite 🔥 plan, you have access to the Cloud Workspace dashboard.

✦ Saving Templates to Cloud

Select any text or prompt in your active editor.
Open the Command Palette (Ctrl+Shift+P / Cmd+Shift+P).
Select Prompteiro: Save Template.
Give it a descriptive name.
The template is immediately synchronized with your cloud account and D1 database.

✦ Accessing Your Saved Templates & History Logs

You can view, search, copy, and manage all your cloud-synced templates and historical code enhancements directly on the web application: 👉 prompteiro.vercel.app/dashboard

✦ Refreshing Entitlement Status

When your plan changes, the status bar will automatically update within 30 seconds.
You can also force a manual refresh instantly by running the Prompteiro: Check Upgrade Status command from the Command Palette.

Help & Troubleshooting

Q: The status bar shows `⚠️ Ollama offline` — how do I fix this?

Make sure the Ollama application is installed and running. Verify at http://localhost:11434 in your browser. Once started, click the warning in the status bar to trigger a reconnect, or trigger any enhancement.

Q: Installation failed with error code `ERR_NET_ENOTFOUND`

This means DNS resolution failed during download. Check your internet connection and firewall settings. If you're behind a corporate proxy, configure proxy settings in VS Code before retrying.

Q: Installation failed with error code `ERR_INSTALLER_EXIT_1` (Windows)

This typically means the UAC (User Account Control) prompt was declined. Re-run the installer and approve the administrator permission dialog when it appears.

Q: I see "Model returned malformed JSON"

This can happen with smaller local models that don't follow the output schema precisely. Try switching to a larger model (e.g., llama3:8b instead of a 1.5B model). This error has been significantly reduced in v0.1.6 with improved JSON extraction and <think> block stripping.

Q: How do I remove or rotate my API keys?

Open the Command Palette → Prompteiro: Setup & Login → navigate to the model switcher → select Remove Key next to the provider. This permanently deletes it from your OS keychain.

Q: What happens when I hit the daily limit?

Free/Spark plan users get 15 local enhancements per day (resets at midnight UTC). Options:

Configure remote mode with your own API key to bypass limits.
Click Upgrade to Ignite in the notification for premium access.

Q: I skipped the initial setup. How do I restart it?

Open the Command Palette (Ctrl+Shift+P) and run Prompteiro: Setup & Login.

In VS Code settings, search for prompteiro — you can reset the consent preference there, or use the Command Palette to reopen the setup wizard.

Changelog

v0.1.6 — Phase 1 Launch (June 2026)

✅ Live download progress bar with percentage and MB/KB display during Ollama install
✅ Structured error codes on all failure paths (ERR_HTTP_STATUS_*, ERR_NET_*, ERR_INSTALLER_*)
✅ Cancellable downloads with abort support (Cancel button during install)
✅ Fixed "Model returned malformed JSON" for reasoning models (DeepSeek-R1, etc.)
✅ Floating toolbar icons now render correctly (VS Code Octicons)
✅ Lenient schema validation — smaller models that omit optional fields still work
✅ Clean build — zero warnings across entire monorepo

v0.1.5

Initial public beta release
Ollama local model setup wizard
Remote API key support (OpenAI, Anthropic, Gemini, Azure)
Floating toolbar with Enhance & Clarify actions
Side-by-side review panel with streaming

Enjoy writing better prompts. ✦

Prompteiro