Memory
Overview
Memory provides persistent context across Claude Code sessions by storing and retrieving project decisions, user preferences, and coding conventions from a local JSON file. When a new session starts, stored memories load automatically so previously established patterns carry forward without re-explanation.
Prerequisites
- Project memory file at
.claude/memories/project_memory.json (created automatically on first memory save)
- Read and Write permissions for the
.claude/memories/ directory
- Claude Never Forgets plugin installed (
/plugin install yldrmahmet/claude-never-forgets)
Instructions
- Access stored memories. On session start, locate and read the memory file at
.claude/memories/project_memory.json using the Read tool. Parse the JSON structure containing timestamped memory entries. See ${CLAUDE_SKILL_DIR}/references/implementation.md for the full retrieval workflow.
- Match memories to current context. Scan memory entries for keywords and topics relevant to the current task. Extract applicable decisions (e.g., "use pnpm instead of npm"), architectural patterns, library choices, and coding style preferences.
- Apply memories silently. Incorporate remembered preferences into responses and tool usage without announcing them. When a memory dictates a package manager, testing framework, or naming convention, follow it automatically.
- Store new memories. When significant decisions occur -- library selections, architectural choices, user-stated preferences, or tool rejections -- write them to the memory file with a timestamp. Add entries via the
/remember command or through automatic capture of conversation signals.
- Resolve conflicts. When a stored memory contradicts a current explicit request, prioritize the current request. Flag the conflict if appropriate and update the memory entry to reflect the new decision. Remove outdated memories that no longer apply.
- Handle cleanup. When memory entries exceed 10, consolidate by removing noise (greetings, acknowledgments) and preserving high-value entries (preferences, decisions, corrections). The cleanup threshold is configurable in
hooks/stop_cleanup.py.
Output
- Automatic loading of stored memories at session start
- Silent application of remembered preferences to tool usage and responses
- New memory entries written with timestamps for significant decisions
- Consolidated memory file after cleanup (preserves important entries, removes noise)
- Explicit memory listing via
/memories command
Error Handling
| Error |
Cause |
Solution |
| Memory file not found |
First session or file deleted |
Initialize a new memory file at .claude/memories/project_memory.json with an empty JSON structure |
| Conflicting memories |
Multiple entries contradict each other |
Apply the most recent memory; suggest cleanup via /forget to remove outdated entries |
| Invalid memory format |
File corrupted or manually edited with syntax errors |
Back up the existing file, recreate with valid JSON structure, restore recoverable entries |
| Permission denied |
File or directory lacks read/write permissions |
Check file permissions on .claude/memories/; request necessary access or use an alternative storage location |
Examples
Automatic preference recall across sessions:
Session 1:
User: "Always use Vitest instead of Jest for this project"
→ Stored to project_memory.json
Session 2:
User: "Add tests for the auth module"
→ Memory loaded: "use Vitest instead of Jest"
→ Test files created with Vitest syntax automatically
Manual memory management:
/remember "This project uses Tailwind CSS v4 with the Vite plugin"
/remember "Deploy to Cloudflare Workers, not Vercel"
/memories # Lists all stored memories with timestamps
/forget "Vercel" # Removes the Vercel-related memory
Tool rejection captured as correction:
set -euo pipefail
User declines a suggested `npm install` action
→ Memory stored: "User prefers pnpm over npm"
→ Future sessions use pnpm automatically
Resources
${CLAUDE_SKILL_DIR}/references/implementation.md -- Step-by-step guide for accessing, applying, updating, and resolving memory conflicts${CLAUDE_SKILL_DIR}/references/errors.md -- Detailed error scenarios with recovery procedures/remember [text] -- Add a new memory entry manually/forget [text] -- Remove a matching memory from storage/memories -- Display all currently stored memories with timestamps
Source: jeremylongshore/claude-code-plugins-plus-skills → skills/.curated/memory/SKILL.md
Also appears in: jeremylongshore/claude-code-plugins-plus-skills/plugins/community/claude-never-forgets/skills/memory/SKILL.md
1---2name: memory3description: 'Execute extract and use project memories from previous sessions for context-aware assistance. Use when recalling past decisions, checking project conventions, or understanding user preferences. Trigger with phrases like "remember when", "like before", or "what was our decision about". '4---5
6# Memory
7
8## Overview
9
10Memory provides persistent context across Claude Code sessions by storing and retrieving project decisions, user preferences, and coding conventions from a local JSON file. When a new session starts, stored memories load automatically so previously established patterns carry forward without re-explanation.
11
12## Prerequisites
13
14- Project memory file at `.claude/memories/project_memory.json` (created automatically on first memory save)
15- Read and Write permissions for the `.claude/memories/` directory
16- Claude Never Forgets plugin installed (`/plugin install yldrmahmet/claude-never-forgets`)
17
18## Instructions
19
201. **Access stored memories.** On session start, locate and read the memory file at `.claude/memories/project_memory.json` using the Read tool. Parse the JSON structure containing timestamped memory entries. See `${CLAUDE_SKILL_DIR}/references/implementation.md` for the full retrieval workflow.
212. **Match memories to current context.** Scan memory entries for keywords and topics relevant to the current task. Extract applicable decisions (e.g., "use pnpm instead of npm"), architectural patterns, library choices, and coding style preferences.
223. **Apply memories silently.** Incorporate remembered preferences into responses and tool usage without announcing them. When a memory dictates a package manager, testing framework, or naming convention, follow it automatically.
234. **Store new memories.** When significant decisions occur -- library selections, architectural choices, user-stated preferences, or tool rejections -- write them to the memory file with a timestamp. Add entries via the `/remember` command or through automatic capture of conversation signals.
245. **Resolve conflicts.** When a stored memory contradicts a current explicit request, prioritize the current request. Flag the conflict if appropriate and update the memory entry to reflect the new decision. Remove outdated memories that no longer apply.
256. **Handle cleanup.** When memory entries exceed 10, consolidate by removing noise (greetings, acknowledgments) and preserving high-value entries (preferences, decisions, corrections). The cleanup threshold is configurable in `hooks/stop_cleanup.py`.
26
27## Output
28
29- Automatic loading of stored memories at session start
30- Silent application of remembered preferences to tool usage and responses
31- New memory entries written with timestamps for significant decisions
32- Consolidated memory file after cleanup (preserves important entries, removes noise)
33- Explicit memory listing via `/memories` command
34
35## Error Handling
36
37| Error | Cause | Solution |
38|-------|-------|----------|
39| Memory file not found | First session or file deleted | Initialize a new memory file at `.claude/memories/project_memory.json` with an empty JSON structure |
40| Conflicting memories | Multiple entries contradict each other | Apply the most recent memory; suggest cleanup via `/forget` to remove outdated entries |
41| Invalid memory format | File corrupted or manually edited with syntax errors | Back up the existing file, recreate with valid JSON structure, restore recoverable entries |
42| Permission denied | File or directory lacks read/write permissions | Check file permissions on `.claude/memories/`; request necessary access or use an alternative storage location |
43
44## Examples
45
46**Automatic preference recall across sessions:**
47
48```
49Session 1:
50User: "Always use Vitest instead of Jest for this project"
51→ Stored to project_memory.json
52
53Session 2:
54User: "Add tests for the auth module"
55→ Memory loaded: "use Vitest instead of Jest"
56→ Test files created with Vitest syntax automatically
57```
58
59**Manual memory management:**
60
61```bash
62/remember "This project uses Tailwind CSS v4 with the Vite plugin"
63/remember "Deploy to Cloudflare Workers, not Vercel"
64/memories # Lists all stored memories with timestamps
65/forget "Vercel" # Removes the Vercel-related memory
66```
67
68**Tool rejection captured as correction:**
69
70```
71set -euo pipefail
72User declines a suggested `npm install` action
73→ Memory stored: "User prefers pnpm over npm"
74→ Future sessions use pnpm automatically
75```
76
77## Resources
78
79- `${CLAUDE_SKILL_DIR}/references/implementation.md` -- Step-by-step guide for accessing, applying, updating, and resolving memory conflicts
80- `${CLAUDE_SKILL_DIR}/references/errors.md` -- Detailed error scenarios with recovery procedures
81- `/remember [text]` -- Add a new memory entry manually
82- `/forget [text]` -- Remove a matching memory from storage
83- `/memories` -- Display all currently stored memories with timestamps
84
85---
86
87**Source:** [`jeremylongshore/claude-code-plugins-plus-skills`](https://github.com/jeremylongshore/claude-code-plugins-plus-skills) → `skills/.curated/memory/SKILL.md`
88
89**Also appears in:** `jeremylongshore/claude-code-plugins-plus-skills/plugins/community/claude-never-forgets/skills/memory/SKILL.md`
thedixitjain