Give your coding agent a way through the tradeoff.

Turn conflicting requirements into evidence-grounded strategies your agent can test. A software-adapted TRIZ CLI and library for agentic development.

Integrate with your agent GitHub Repository

WORKFLOW.DIAGRAM

1. Request & Evidence

2. Contradiction Analysis

3. Candidate Strategy (JSON)

4. Implementation & Checks

01 / Process

Agentic Coding Workflow

Clearly separate strategy analysis from implementation execution. The tool does not write code or run tests by itself.

Manager supplies evidence

You define the architectural requirement conflict (e.g., fresher dashboards vs database load).

Agent analyzes via TRIZ

The agent invokes the TRIZ CLI to classify the contradiction and generate deterministic strategy mappings.

$ triz analyze request.json --offline response.json --out result.json

JSON carries classification

Persisted JSON output retains evidence bounding, catalog versions, and candidate principles.

$ triz inspect result.json

{
  "classification": "technical",
  "evidenceIds": ["req-42"],
  "catalogVersion": "triz-software-2",
  "inferenceDeterministic": false
}

Manager chooses & Agent implements

The manager reviews the hypothesis, and the agent implements the chosen candidate with independent checks.

02 / Toolbelt

Core Capabilities

Evidence Validation

Uncertain or insufficient results trigger a handback for more evidence rather than fabricating confidence.

Persisted JSON

Offline fixture analysis maintains typed, evidence-bound classifications across sessions. No browser service needed.

40 Principles & 6 Routes

A refined toolbelt of software-adapted inventive principles and physical contradiction routes.

03 / Integration

Setup & Install

Requires Node 24+. Portable compiled JavaScript with no runtime npm dependencies. Optional live Jev integration via Pi SDK. Junior integration is optional for orchestration.

Give your coding agent TRIZ

For Cursor, Codex, Claude Code, Copilot, Gemini CLI, and other agents with terminal access.

Set up TRIZ for this project. Read https://trizit.dev/INSTALL.md and follow the instructions. Install the skill, wire up managed agent instructions, verify offline readiness, and report anything needing my attention before live analysis.

Agent setup guide Install script TRIZ skill

Install from Source

$ git clone https://github.com/matt-cochran/triz.git
$ cd triz
$ npm ci
$ npm run build
$ node dist/triz.js catalog

Note: GitHub release tarballs install without compilation. There is no npm registry publication, IDE plugin, or MCP server.

A toolbelt for software contradictions.

Explore definitions, inventive principles, and physical resolution routes from the software-adapted catalog.

Search the contradiction library →

What is TRIZ?

TRIZit adapts selected TRIZ principles and contradiction-resolution routes for software architecture, implementation tradeoffs, and agentic coding. Your agent uses supplied evidence to frame the conflict, explore strategies, and return a reviewable handoff.

TRIZ—the theory of inventive problem solving—is a systematic methodology developed from recurring patterns in inventions and patent literature. It includes tools for resolving contradictions, analyzing systems, and guiding inventive solutions. TRIZ background →

Technical Contradiction

Improving one outcome worsens another (e.g., speed vs accuracy).

Physical Contradiction

The same element needs opposing properties at different times or places.