calesthio

SessionAnchor — Development skill for Claude Code

Development community

One-command context memory for Claude Code sessions.

How to install SessionAnchor

This entry records only its repository, not the path inside it, so there is no exact command to give. Open calesthio/SessionAnchor and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What SessionAnchor does

One-command context memory for Claude Code sessions. Local SQLite, zero dependencies.

Alternatives in Development

Sanitize — Detect and redact PII from text files — 15 categories (SSNs, credit cards, API keys, etc.), zero dependencies 3.1k ★
Claw Code Agent — Claw Code No Rust No TypeScript Only Python 543 ★
Memstack Search — Search MemStack memory for past sessions, insights, and project context 294 ★

README

SessionAnchor

[![PyPI version](https://img.shields.io/pypi/v/sessionanchor.svg)](https://pypi.org/project/SessionAnchor/) [![Python](https://img.shields.io/pypi/pyversions/sessionanchor.svg)](https://pypi.org/project/SessionAnchor/) [![License](https://img.shields.io/badge/license-AGPL%20v3-blue.svg)](https://github.com/calesthio/SessionAnchor/blob/master/LICENSE)

One-command context memory for [Claude Code](https://docs.anthropic.com/en/docs/claude-code) sessions.

The Problem

Claude Code sessions are stateless. Every new conversation starts cold. Claude re-reads your entire `CLAUDE.md`, architecture docs, and context files from scratch, burning through tokens and losing the decisions, progress, and institutional knowledge from your last session.

If your project has grown any real context (and most do), you're paying for it on every single session start:

A typical project accumulates 10-30KB of context files
That's ~15,000-25,000 tokens loaded every session, most of it stale
Decisions made three weeks ago get equal weight to yesterday's blockers
There's no continuity: Claude doesn't know what it did last time

The Solution

`SessionAnchor` replaces flat context files with a local SQLite memory store that uses tiered retrieval. One command gives your repo continuity across sessions.

Before: ~18,000 tokens loaded from markdown files every session
After:   ~2,000 tokens loaded from structured memory (9x reduction)

The boot briefing is compact, current, and structured. Claude gets exactly what it needs to resume work, not a wall of historical text.

Who This Is For

This tool is built for a specific scenario:

You use Claude Code (Anthropic's CLI agent) on real projects
Your projects have accumulated context that Claude needs across sessions
You want local-only, zero-setup continuity without hosted services or API keys
You care about token efficiency and want to measure it

It is **not** a general-purpose memory