sandip124 avatar

Wisegit

sandip124/wisegit

Turns git history into a protection layer for AI code edits. Before Claude modifies a file, it calls get_file_decisions to see which functions are frozen, stable, or open based on commit intent analysis. It parses your repo's commits with Tree-sitter, extracts decision context from messages and linked issues, then assigns freeze scores to individual functions. A manually tested sleep() for a race condition gets marked FROZEN so the AI won't remove it as dead code. Setup is one command: npx @sandip124/wisegit setup indexes your history into local SQLite and drops a .mcp.json for auto-discovery. You get ten MCP tools including override creation, impact prediction, and convention extraction. Works entirely locally with no external services.

GitHub stars
1
tools
11
transport
STDIO
last updated
Mar 26, 2026

Tools

11 tools

Public tool metadata for what this MCP can expose to an agent.

get_file_decisionsDecision manifest for a file — freeze scores, intent history, recovery levels, override status

Decision manifest for a file — freeze scores, intent history, recovery levels, override status

No parameter schema in public metadata yet.

get_freeze_scoreScore + signal breakdown for a specific function

Score + signal breakdown for a specific function

No parameter schema in public metadata yet.

get_function_historyFull chronological decision timeline for a function

Full chronological decision timeline for a function

No parameter schema in public metadata yet.

get_theory_gapsFunctions with unrecoverable rationale (inactive authors, timeline gaps)

Functions with unrecoverable rationale (inactive authors, timeline gaps)

No parameter schema in public metadata yet.

get_branch_contextBranch merge history — what was migrated and why

Branch merge history — what was migrated and why

No parameter schema in public metadata yet.

search_decisionsSearch past decisions by keyword across the entire repo

Search past decisions by keyword across the entire repo

No parameter schema in public metadata yet.

create_overrideOverride a frozen function (user approves in Claude Code UI)

Override a frozen function (user approves in Claude Code UI)

No parameter schema in public metadata yet.

extract_intentExtract intent for NOISE commits using the host LLM — no Ollama needed

Extract intent for NOISE commits using the host LLM — no Ollama needed

No parameter schema in public metadata yet.

find_similar_functionsSearch for existing functions that solve a similar problem before writing new code

Search for existing functions that solve a similar problem before writing new code

No parameter schema in public metadata yet.

predict_impactPredict what functions will break if a given function is modified

Predict what functions will break if a given function is modified

No parameter schema in public metadata yet.

get_codebase_conventionsExtract coding conventions for a file's neighborhood

Extract coding conventions for a file's neighborhood

No parameter schema in public metadata yet.

Sponsors

Readme

wisegit

npm MCP Registry License: MIT

"Don't take a fence down until you know the reason it was put up." — G.K. Chesterton

wisegit is a local MCP server that extracts decision intent from git history and protects intentional code from AI modification.

When Claude Code (or any MCP-compatible agent) is about to edit a file, wisegit injects a decision manifest showing which functions are frozen, stable, or open — so the AI respects what was intentional, not just what compiles.

Zero config. Zero external services. Everything local.

Install

# Set up any repo (one command)
npx @sandip124/wisegit setup

# Or add as MCP server globally
claude mcp add wisegit -- npx @sandip124/wisegit serve

Published on


View the full README on GitHub

Details

Repository
github.com/sandip124/wisegit
Category
Developer Tools
Package
@sandip124/wisegit
Transport
STDIO
Registry
MCP registry, active

Related Developer Tools MCP servers

View all
1sanoy24 avatarrepowardsanoy24/repowardGitHub MCP server that blocks private-to-public leaks by agents, with approvals and audit.1stars
2sean-sunagaku avatarPromptlint MCPsean-sunagaku/promptlint-mcpStatic linter for AI prompts: contradictions, redundancy, ambiguity, and fluff.1stars
3tencent-lexiang avatarDEPRECATEDtencent-lexiang/lexiang-mcp-serverDEPRECATED: Please use io.github.tencent-lexiang/lexiang-mcp-server instead.1stars
4yksanjo avatarGmemyksanjo/gmemPersistent, Solana-aware project memory for AI coding agents. Typed entities + git-aware diffs.1stars
5yubinkim444 avatarRepo Memoryyubinkim444/repo-memoryShared, git-tracked working memory for AI agents on the same codebase.1stars
6agentic-eng avatarA2atlassianagentic-eng/a2atlassianJira & Confluence for AI agents. Pre-configured connections, compact output.0stars