SpecLoreProduct-Engineering Collaboration for the AI Coding Era

Turn requirements into verifiable BDD specs, and acceptance testing into an automated pipeline

Get Started

GitHub

Install in Seconds

$ pnpm add -g speclore

From Requirements to Verification, One Pipeline

📄

Requirements

Markdown / Word / URL / Plain text

→

🧪

BDD .feature

Standard Gherkin acceptance criteria

→

🤖

Constraints + Scaffolding

AI coding constraints + test scaffolding

→

✅

Verification Report

Auto test mapping + acceptance results

Multi-Format Requirement Ingestion

Supports Markdown, Word, Excel, PDF, image OCR, URL, and plain text — any format of requirement input is unified into standard BDD .feature acceptance criteria.

speclore spec requirements.md    # Markdown
speclore spec design.docx        # Word
speclore spec mockup.png         # Image OCR
speclore spec "reset password"   # Plain text

AI Coding Constraint Generation

Auto-detects Cursor / Claude Code / Qoder in your project, generates coding constraint files and test scaffolding in the correct format — AI coding automatically follows business rules.

speclore code
# Output:
# .cursor/rules/speclore.mdc
# .claude/rules/speclore.md
# .qoder/rules/speclore.md
# tests/**/*.test.ts (test scaffolding)

Automated Acceptance Mapping

After running tests, results are automatically mapped back to .feature scenarios. Three mapping strategies: mapping files (auto), explicit markers (manual), pattern matching (config).

speclore verify
# ✅ 5/5 scenarios passed (100%)
#
# specs/order/create.feature
#   ✓ Create valid order
#   ✓ Reject when inventory insufficient
#   ✓ Warn on duplicate items

MCP-Native Integration

4 MCP tools exposed via standard protocol, directly callable by AI clients. Each call returns workflow state and next-step guidance — out-of-order operations automatically error.

# AI clients call via MCP automatically:
speclore.status  → Check status
speclore.spec    → Generate feature
speclore.code    → Generate constraints
speclore.verify  → Acceptance test

Works with Leading AI Coding Clients

speclore setup auto-detects and configures MCP — no manual setup needed

Cursor

Claude Code

Qoder

Full Workflow Demo

speclore

$ speclore setup

✔ Detected AI tools: Cursor, Qoder
✔ Written .cursor/mcp.json
✔ Written .qoder-cn/mcp.json
✔ Generated .speclore/config.yaml

Zero to Acceptance in Three Steps

bash

# 1. Initialize your project
cd your-project && speclore setup

# 2. Generate .feature acceptance criteria from requirements
speclore spec "Patient registration requires phone verification"

# 3. Generate AI coding constraints + test scaffolding
speclore code

Code in your AI client, then run acceptance: