no-cursor (nc)

A local, CLI-first code assistant designed for a private, fast, and deterministic experience. No cloud APIs, no background agents, no latency—just pure local intelligence focused on your active file.

Why no-cursor?

Proactive Auditing

The new fix command actively audits your file for logic bugs and security flaws using local LLM reasoning.

Strategic Planning

Generate step-by-step roadmaps for your complex feature goals with the plan and save-plan system.

Semantic Search

Find logic and patterns within your file using natural language, powered by local semantic indexing.

Interactive Assistance

Live, file-aware chat mode with full history persistence stored directly in your workspace.

Deterministic Edits

Changes are shown as resilient unified diffs with built-in confidence scoring and hash verification.

Air-Gapped Private

Runs entirely on your machine. Zero usage tracking. Zero cloud dependencies. You own every byte.

The Assistant Console

Typical NC Assistant workflow:

nc (test.py)> chat
Entering Interactive Chat for test.py. Type 'exit' to return.
nc (chat:test.py)> how does the divide function work?
Assistant: The divide function returns b / a. It seems the arguments are reversed...

nc (test.py)> fix
Auditing file for bugs...
[Bug Audit & Fix Suggestion]
The divide function is logically reversed (b/a instead of a/b).
--- a/test.py
+++ b/test.py
@@ -3,4 +3,4 @@
 def divide(a, b):
-    return b / a
+    return a / b

Apply detected fix? [y/n]: y
✓ Applied. Backup saved to .nc/backup/

nc (test.py)> plan "Add logging and exception handling"
[Implementation Plan]
1. [ ] Wrap b/a in a try-except block
2. [ ] Import logging module
3. [ ] Add log.info for results

Plan generated. Use 'save-plan' to keep it.
nc (test.py)> save-plan
✓ Plan for 'Add logging...' saved to workspace.

nc (test.py)> stats
Usage: 1.2k tokens | Total Time: 45s | Calls: 12

Command Reference

open <file>

Focus file for editing

edit <msg>

Request code changes

chat <msg>

Live contextual assistance

fix

Audit file for bugs

plan <goal>

Generate feature roadmap

search <msg>

Semantic search in file

apply

Commit changes to disk

revert

Restore last backup

stats

Show usage statistics

Status & Roadmap

Assistant ecosystem ready — Core tools for editing, auditing, and searching are fully functional.

Next priorities:

Multi-file context injection
Sequential plan execution automation
Conversation summary for long histories
Workspace-wide semantic indexing

Guaranteed Principles

Unifile focus — Total context for the file you are actually touching.
Inspectable diffs — Human-readable changes with confidence scores.
State Persistence — File-specific chat history and saved plans.
Deterministic — Local GGUF models mean reliable, repeatable results.