HackStack

Project Info

ToneBridge

Devpost

Inspiration

We’ve all sent messages that were misinterpreted texts read as “angry” because they lacked context, emails seen as “aggressive” because they were brief. In our digital-first world, we’re losing the nuance of human connection. ToneBridge was built to stop this friction before it starts, bridging the gap between intent and understanding.

What it does

ToneBridge is a Human Interaction Mediator powered by Gemini AI. It doesn’t just check grammar it reads for emotional tone and helps people communicate better. Analyzes your message for risk (e.g., “High Risk: Accusatory”) Explains why it might be misread (e.g., “This sounds defensive because…”) Rewrites it instantly in three human-centered tones: Calm & Friendly – for casual connection Professional & Clear – for workplace clarity Empathetic & Supportive – for deepening relationships How I built it Frontend: HTML, CSS and JavaScript for a lightweight, intuitive interface Backend: PHP with PDO for secure API calls AI Engine: Google Gemini (gemini-2.5-flash) with a carefully engineered system prompt Prompt Engineering: 80% of effort focused on making Gemini act as a “Mediator,” with strict JSON output for parsing Storage: MySQL logs interactions for future analytics The result is fast, reliable and human-centered. Challenges I ran into Token limits: Needed to increase maxOutputTokens to ensure full three-message rewrites Parsing consistency: LLM responses required strict JSON schemas to avoid ambiguity Emotional nuance: Balancing professional vs robotic tone took multiple iterations Accomplishments that I'm proud of Zero-friction UX: Fully functional in under 5 seconds, no login required Human-centered design: Every element, from explanations to colors, reduces stress Real-time AI analysis: Three thoughtful rewrites returned nearly instantly What I learned Prompting is programming: AI behavior depends entirely on how its role is defined UX is emotional: Users under stress need a calming interface, not a “techy” one

What's next

Browser Extension: Fix messages directly in Gmail or Slack Tone Slider: Users can control the level of formality Multi-language Support: Bridging cultural gaps and tonal differences

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSClaimed
  • Google GeminiClaimed
  • HTMLClaimed
  • JavaScriptClaimed
  • PHPClaimed
  • SQLClaimed

0 of 6 appear in the indexed code. 6 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

1.2 KB

Source files

1

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars