Work with local models
Use Ollama, LM Studio or an optional loopback OpenAI-compatible endpoint with capability-specific model routing.
Active validation
A local AI coding workspace built around reviewable engineering, not blind automation.
CloneCode brings local models, repository-aware context, code intelligence, Git review and governed agent execution into a native Windows workspace.
CLONECODE · LOCAL AI WORKSPACE
src/
core/
ui/
tests/
README.md
pyproject.toml
core/
ui/
tests/
README.md
pyproject.toml
def verify(change):
tests = run_tests()
if tests.green:
return review(change)
# local model route
mode = "Agent"
tests = run_tests()
if tests.green:
return review(change)
# local model route
mode = "Agent"
LOCAL AI
Plan prepared.
RED evidence captured.
Implementation bounded.
Fresh verification pending.
Plan prepared.
RED evidence captured.
Implementation bounded.
Fresh verification pending.
CloneCode · current product-surface preview
What it does
Built around the work users actually need to finish.
Current capabilities are described in plain language and separated from future scope.
Code in one workbench
Files, sessions, agents, Git, models, skills, connectors, editor and assistant live in one Windows-first interface.
Choose the authority level
Ask, Plan, Agent and Sandbox Agent modes keep conversational help separate from workspace-changing actions.
Use real code intelligence
LSP diagnostics, navigation, formatting, syntax color, inline completion and repository-aware sparse context support real codebases.
Review Git changes
Inspect status, diffs, branches, staging, commits and per-hunk stage/reject operations before accepting work.
Verify before completion
The governed coding path uses test-first RED/GREEN evidence, independent review and fresh verification before declaring success.
CloneCode is in active validation. Windows 11 x64 is the supported release lane; release certification and host-specific local-model performance must be measured rather than inferred.