pysince

MCP server for coding agents: detects stale file reads before your agent acts on outdated content. Zero dependencies.

pip install pysince

What it does

pysince is a local MCP server that stamps every file your agent reads with its mtime and SHA-256 hash. On any subsequent tool call, it compares stored fingerprints against the current file and surfaces all tracked files that have changed, unprompted. The agent does not need to remember to check anything.

Files changed since last read:
  config.json (content changed, mtime changed) read 4m ago (+2 lines)
  alerts.py (content changed, mtime changed) read 3m ago

How it works

Every file the agent reads gets stamped with its current mtime and content hash. On any later MCP call, pysince checks all tracked files: mtime first (fast os.stat), full hash only if mtime moved. No daemon, no polling, no background process. Just a comparison against disk at the next turn.

Setup

Add to your MCP client config:

VS Code (.vscode/mcp.json):

{
  "servers": {
    "pysince": {
      "type": "stdio",
      "command": "pysince-mcp"
    }
  }
}

Antigravity (mcp_config.json):

{
  "mcpServers": {
    "pysince": {
      "command": "python",
      "args": ["-m", "since.mcp"]
    }
  }
}

Then add this line to your agent's system instructions:

On the first read of any file, call stamp_file_read. Before editing a file, call check_staleness. When the response lists changed files, re-read them before acting on their contents.

Tools

check_staleness never answers only about the one file asked. It reports the full set of tracked files that have drifted, so the agent cannot stay blind to a change it did not think to check.

When to use

Strongest when files change outside the agent's view: another process, a teammate on the same repo, a formatter, a pre-commit hook, a parallel agent, or context that drifted over a long session.

Less useful for quick single-file edits an agent already re-reads on its own. Skip it for throwaway scripts. Reach for it when agents share files, sessions run long, or more than one actor touches the tree.

Chat apps

pysince also provides a @since_time decorator for wrapping LLM chat functions with temporal context: timestamps, session duration tracking, gap detection, and stale context surfacing.

from since import Store, since_time
from openai import OpenAI

store = Store("~/.since/chat.db")
client = OpenAI()

@since_time(store=store, timezone="Asia/Kolkata")
def chat(messages):
    return client.chat.completions.create(model="gpt-4o-mini", messages=messages)

resp = chat(messages=[{"role": "user", "content": "hello"}])
print(resp.choices[0].message.content)

The model receives a time block before each turn:

Now:      Wed Jul 01, 02:36 AM (night)
Session:  9h 2m total, 4m active across 3 sittings, 8 messages
Gap:      6h since the last message

Works with OpenAI, Anthropic, and Gemini response shapes automatically. Pass extract_reply= for other providers.

CLI

pysince enrich <session_id>   # Enrich a session with temporal context
pysince search <session_id>   # Search messages by time range
pysince stats <session_id>    # Show session statistics

Requirements

• Python 3.10+
• Zero dependencies

The PyPI name is pysince because since was already taken. The import name and repo name are since.

Links

• GitHub
• PyPI