How guhanavish/mirrormap works
MirrorMap is a read-only project intelligence and spec-drift auditing tool. Its main workflow takes requirements, a repository inventory, and optional reviewed verification evidence; it compares those inputs, classifies findings, and produces reports and a human-review remediation plan. The offline atlas presents the results, while JUnit import, route discovery, agent-skill access, and MCP provide distinct supporting capabilities. The sampled source shows key module references; other capability links rely on the README, and unsampled internals are not inferred.
Main entry points: experience.py Audit Workflow · core.py Evidence Comparison · evidence.py Test Observations · data-model.md Approved Requirements
Components 17
Audit Inputs
- Approved Requirements
references/data-model.md - Target Repository
- Test Observations
projectmirror/evidence.py
Audit Engine
- Audit Workflow
projectmirror/experience.py - Source Inventory
projectmirror/core.py - Route Candidates
projectmirror/frameworks.py - Evidence Comparison
projectmirror/core.py - Review-Only Plan
projectmirror/core.py
Results and Evidence
- JUnit Import
projectmirror/evidence.py - Audit Reports
projectmirror/experience.py - Interactive Atlas
projectmirror/dashboard.py
User Interfaces
- CLI Entry
projectmirror/__main__.py
Integrations
- Agent Skill
SKILL.md - GitHub Action
action.yml - MCP Server
projectmirror/mcp.py
Other components
- Engineer or Agent
- External Coding Agent
Connections 21
- Engineer or Agent → CLI Entry: runs commands
projectmirror/__main__.py - CLI Entry → Audit Workflow: dispatches to
projectmirror/__main__.py - Engineer or Agent → Approved Requirements: supplies
README.md - Engineer or Agent → Target Repository: selects
README.md - Audit Workflow → Source Inventory: indexes repository
README.md - Source Inventory → Route Candidates: discovers candidates
projectmirror/core.py - Audit Workflow → Evidence Comparison: compares evidence
README.md - Approved Requirements → Evidence Comparison: defines intent
README.md - Target Repository → Source Inventory: scans source
README.md - Engineer or Agent → Test Observations: provides evidence
README.md - Audit Workflow → JUnit Import: imports reports
projectmirror/experience.py - JUnit Import → Test Observations: produces observations
README.md - Test Observations → Evidence Comparison: informs findings
README.md - Evidence Comparison → Review-Only Plan: creates proposals
README.md - Audit Workflow → Audit Reports: writes reports
README.md - Audit Workflow → Interactive Atlas: renders dashboard
projectmirror/experience.py - Engineer or Agent → Agent Skill: installs or uses
README.md - Engineer or Agent → GitHub Action: invokes audit
README.md - Engineer or Agent → MCP Server: connects through
README.md - MCP Server → Audit Workflow: invokes audit
projectmirror/mcp.py - Audit Reports → External Coding Agent: guides approved work
README.md