Skip to content

Releases: severity1/claude-code-auto-memory

v0.8.0

13 Dec 08:51

Choose a tag to compare

Changes

Template Enforcement

  • Enforce template structure in codebase-analyzer skill for consistent CLAUDE.md generation
  • Add explicit marker syntax rules (no BEGIN prefix, no name in closing tag)
  • Add shared guidelines reference file

Marketplace Rename

  • Rename marketplace from claude-code-marketplace to severity1-marketplace

CLAUDE.md Improvements

  • Add auto-generated CLAUDE.md with proper template markers
  • Document configuration options (trigger modes)
  • Document inline commit context pattern

Documentation

  • Fix README.md installation instructions
  • Add SKILL.md references to templates

Full Changelog

v0.7.0...v0.8.0

v0.7.0

05 Dec 08:56

Choose a tag to compare

What's New

Skill Improvements

  • Better trigger conditions: Skill descriptions now include "Use when..." context for better invocation
  • Shared guidelines: Extracted common Claude Code memory guidelines to skills/shared/references/guidelines.md (reduces duplication by ~70 lines)
  • Consolidated definitions: Section types defined once in authoritative tables, referenced elsewhere

Content Verification

  • Removal verification: Before removing documented content, uses Grep to search entire codebase - keeps if found elsewhere, removes only if truly gone
  • Stale command detection: Compares documented commands against successfully executed commands and updates to match what worked (e.g., python pytest -> python -m pytest)
  • Conventions vs patterns: Clear distinction - conventions are explicit human rules, patterns are AI-detected recurring structures

Subtree Memory Support

  • scripts/CLAUDE.md: Hook script patterns and conventions
  • tests/CLAUDE.md: pytest testing conventions and patterns

Bug Fixes

  • Fixed dirty-files path in status command (.claude/.dirty-files -> .claude/auto-memory/dirty-files)

Agent Updates

  • Memory-updater agent now includes Grep tool for pattern verification

Full Changelog

v0.6.0...v0.7.0

v0.6.0

03 Dec 12:20

Choose a tag to compare

What's New

Inline Commit Context (#6)

  • Commit hash and message are now embedded inline in dirty-files: /path/to/file [hash: message]
  • Eliminates separate commit-context.json file
  • Deduplication logic updates existing entries with commit context instead of creating duplicates

Configurable Trigger Modes (#6)

  • default mode: Tracks all Edit/Write/Bash operations immediately
  • gitmode: Only tracks git commits (cleaner, delayed)
  • Configure via /auto-memory:init command or .claude/auto-memory/config.json

New /auto-memory:sync Command (#5)

  • Detect and process manually edited files without full recalibration
  • Uses git diff to find changes made outside Claude Code

Bug Fixes

  • Switch memory-updater agent from haiku to sonnet (fixes #7 - haiku doesn't support extended thinking)
  • Fix git commit detection for chained commands (git add && git commit)

Full Changelog

v0.5.0...v0.6.0

v0.5.0

01 Dec 08:55

Choose a tag to compare

What's New

Instant Memory Refresh

The Stop hook now instructs Claude to read CLAUDE.md after the memory-updater agent completes, so your main session has fresh memory immediately. No more waiting until next session to benefit from updated project context.

Python 3.8 Compatibility Fix

Fixed TypeError: 'type' object is not subscriptable error that some users were hitting. Added from __future__ import annotations for compatibility with mixed Python environments.

Terminology Standardization

Standardized on "memory" terminology throughout (matching Claude Code official docs and the plugin name).

Full Changelog

v0.4.0...v0.5.0

v0.4.0

30 Nov 06:08

Choose a tag to compare

Changes

  • Remove memory- prefix from command files
    • memory-init.md -> init.md
    • memory-calibrate.md -> calibrate.md
    • memory-status.md -> status.md
  • Commands now use /auto-memory:{init|calibrate|status} namespace
  • Update documentation to reflect new command structure

Full Changelog

v0.3.0...v0.4.0

v0.3.0

30 Nov 05:50

Choose a tag to compare

Changes

  • Standardized CLAUDE.md terminology as memory files (not documentation)
  • Updated tagline from "Zero tokens" to "Minimal tokens" for accuracy
  • Consistent memory terminology across README, skills, commands, and config
  • Commands use filename-based auto-detection (no explicit name in frontmatter)

Full Changelog

v0.2.0...v0.3.0

v0.1.0 - Initial Release

29 Nov 05:31

Choose a tag to compare

auto-memory v0.1.0

Your CLAUDE.md, always in sync. Minimal tokens. Zero config. Just works.

A Claude Code plugin that automatically maintains CLAUDE.md files as your codebase evolves.

Features

  • Automatic sync: Tracks Edit/Write/Bash operations and updates CLAUDE.md at end of turn
  • Bash operation tracking: Detects rm, mv, git rm, git mv, unlink commands
  • Minimal-token tracking: PostToolUse hook has no output; stop hook triggers isolated agent
  • Isolated processing: Agent runs in separate context window, doesn't consume main session tokens
  • Marker-based updates: Only modifies AUTO-MANAGED sections, preserves manual content
  • Subtree support: Hierarchical CLAUDE.md for monorepos

Commands

  • /memory-init - Initialize CLAUDE.md structure with interactive wizard
  • /memory-calibrate - Force full recalibration
  • /memory-status - Show sync status

Installation

claude plugin marketplace add severity1/claude-code-marketplace
claude plugin install auto-memory@claude-code-marketplace