Local, git-style memory for AI agents (MCP server).
Drift inferred · capture-to-capture
No drift recorded — single capability capture; advisories appear once its surface changes.
transport stdio · http counts 6 tools · 0 res
· 0 prompts
permission surface via code analysis
tools
-
commit_turn
Journal a finished turn; durable phrasing auto-merges into the wiki
-
context_stats
Token meter: compilations, tokens served, tokens saved
-
explain_selection
Per-item salience scores + exclusion reasons for a prompt
-
full_context
Page through the complete raw history (token counts included)
-
prepare_context
Compile a token-budgeted context patch relevant to the prompt, with token accounting
-
search_context
BM25 search over all events + wiki pages
evidence-backed
findings quoted directly from the published source artifact — not inferred
filesystem 4
- fs RT123-new-contextgit-mcp-7e0d9f5/src/contextgit/cli.py :283
with open(args.out, "w", encoding="utf-8") as f: - fs RT123-new-contextgit-mcp-7e0d9f5/src/contextgit/core/store.py :37
with open(path, "r", encoding="utf-8") as f: - fs RT123-new-contextgit-mcp-7e0d9f5/src/contextgit/install.py :14
import shutil - fs RT123-new-contextgit-mcp-7e0d9f5/src/contextgit/usage.py :50
with open(self.path, "a", encoding="utf-8") as f:
network 1
- net RT123-new-contextgit-mcp-7e0d9f5/src/contextgit/ui.py :25
from urllib.parse import parse_qs, urlparse
declared dependencies 1
- pydantic@>=2.0.0