jeremylongshore/tons-of-skills-marketplace / plugins/community/claude-never-forgets/skills/memory
memory
VerifiedExecute 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".
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.0 KB |
| references/errors.md | markdown | 890 B |
| references/examples.md | markdown | 7.2 KB |
| references/implementation.md | markdown | 1.4 KB |
| scripts/manage-memory.pyexecutable | script | 4.6 KB |
references/examples.md
View on GitHub# Examples
## Example 1: Auto-Loading Memories at Session Start
When a new session begins, the memory skill reads the stored memory file and applies
remembered preferences silently.
**Memory file** (`.claude/memories/project_memory.json`):
```json
{
"memories": [
{
"id": "m-001",
"text": "Use pnpm as the package manager, never npm or yarn",
"timestamp": "2026-03-10T14:22:00Z",
"source": "user-correction"
},
{
"id": "m-002",
"text": "Project uses Vitest for testing, not Jest",
"timestamp": "2026-03-10T15:05:00Z",
"source": "user-preference"
},
{
"id": "m-003",
"text": "Deploy target is Cloudflare Workers via wrangler CLI",
"timestamp": "2026-03-11T09:30:00Z",
"source": "explicit-remember"
},
{
"id": "m-004",
"text": "Prefer functional components with composables over class-based patterns",
"timestamp": "2026-03-12T11:00:00Z",
"source": "user-preference"
}
]
}
```
**Session behavior**:
```
Session start →
Read .claude/memories/project_memory.json
Parse 4 memory entries
Apply silently:
- Package manager: pnpm (m-001)
- Test runner: vitest (m-002)
- Deploy target: Cloudflare Workers (m-003)
- Code style: functional + composables (m-004)
User: "Add tests for the auth module"
→ Creates auth.test.ts using vitest syntax (imports from 'vitest', uses describe/it/expect)
→ Runs with: pnpm test -- --grep auth
→ No mention of Jest or npm — preferences applied silently
```
## Example 2: Storing New Memories with /remember
The `/remember` command writes a new timestamped entry to the memory file.
```bash
# Store a library preference
/remember "Use zod for all runtime validation, not joi or yup"
# Store an architectural decision
/remember "API routes follow /api/v1/{resource} convention with kebab-case"
# Store a tool preference
/remember "Use biome for formatting instead of prettier + eslint"
```
**Resulting memory file update**:
```json
{
"id": "m-005",
"text": "Use zod for all runtime validation, not joi or yup",
"timestamp": "2026-03-15T10:15:00Z",
"source": "explicit-remember"
}
```
**Verification**:
```bash
/memories
# Output:
# 1. [2026-03-10] Use pnpm as the package manager, never npm or yarn
# 2. [2026-03-10] Project uses Vitest for testing, not Jest
# 3. [2026-03-11] Deploy target is Cloudflare Workers via wrangler CLI
# 4. [2026-03-12] Prefer functional components with composables over class-based patterns
# 5. [2026-03-15] Use zod for all runtime validation, not joi or yup
```
## Example 3: Implicit Memory Capture from Corrections
When the user corrects a tool choice or approach, the correction is captured
as a memory without needing an explicit `/remember` command.
```
User: "Install the testing library"
Agent: runs `npm install vitest`
User: "No, I said use pnpm"
Agent:
→ Corrects to: pnpm install vitest
→ Stores memory: "User prefers pnpm over npm" (source: user-correction)
→ Future sessions automatically use pnpm
```
**Another implicit capture**:
```
User: "Create a new component"
Agent: Creates a class-based component
User: "We use functional style with setup() here"
Agent:
→ Rewrites as functional component
→ Stores memory: "Prefer functional components with setup() over class-based"
→ All future component creation uses functional style
```
## Example 4: Resolving Conflicting Memories
When a stored memory contradicts a current explicit instruction, the current
instruction wins and the memory is updated.
```
Stored memory: "Deploy target is Cloudflare Workers"
User: "We switched to Vercel, deploy there instead"
Agent behavior:
1. Detects conflict: stored memory says Cloudflare, user says Vercel
2. Prioritizes current explicit instruction (Vercel)
3. Updates memory file:
- Removes: "Deploy target is Cloudflare Workers"
- Adds: "Deploy target is Vercel (changed from Cloudflare Workers)"
4. Proceeds with Vercel deployment
```
## Example 5: Memory Cleanup When Entries Exceed Threshold
When the memory file exceeds 10 entries, low-value entries are removed.
**Before cleanup** (12 entries):
```json
{
"memories": [
{"id": "m-001", "text": "Use pnpm", "source": "user-correction"},
{"id": "m-002", "text": "Use Vitest", "source": "user-preference"},
{"id": "m-003", "text": "Deploy to Vercel", "source": "explicit-remember"},
{"id": "m-004", "text": "Functional components", "source": "user-preference"},
{"id": "m-005", "text": "Use zod for validation", "source": "explicit-remember"},
{"id": "m-006", "text": "User said hello", "source": "auto-capture"},
{"id": "m-007", "text": "User acknowledged the change", "source": "auto-capture"},
{"id": "m-008", "text": "API uses /api/v1/ prefix", "source": "explicit-remember"},
{"id": "m-009", "text": "Use biome for formatting", "source": "explicit-remember"},
{"id": "m-010", "text": "User said thanks", "source": "auto-capture"},
{"id": "m-011", "text": "Database is PostgreSQL via Drizzle ORM", "source": "user-preference"},
{"id": "m-012", "text": "Auth uses Lucia v3", "source": "explicit-remember"}
]
}
```
**After cleanup** (noise removed, high-value kept):
```json
{
"memories": [
{"id": "m-001", "text": "Use pnpm", "source": "user-correction"},
{"id": "m-002", "text": "Use Vitest", "source": "user-preference"},
{"id": "m-003", "text": "Deploy to Vercel", "source": "explicit-remember"},
{"id": "m-004", "text": "Functional components", "source": "user-preference"},
{"id": "m-005", "text": "Use zod for validation", "source": "explicit-remember"},
{"id": "m-008", "text": "API uses /api/v1/ prefix", "source": "explicit-remember"},
{"id": "m-009", "text": "Use biome for formatting", "source": "explicit-remember"},
{"id": "m-011", "text": "Database is PostgreSQL via Drizzle ORM", "source": "user-preference"},
{"id": "m-012", "text": "Auth uses Lucia v3", "source": "explicit-remember"}
]
}
```
Removed entries m-006, m-007, m-010 (greetings, acknowledgments, thanks) as noise.
## Example 6: Forgetting a Memory with /forget
Remove a specific memory when it is no longer relevant.
```bash
/forget "Cloudflare"
# Searches memory entries for "Cloudflare"
# Removes: "Deploy target is Cloudflare Workers"
# Confirms: "Removed 1 memory matching 'Cloudflare'"
```
```bash
/forget "Jest"
# If no match found:
# "No memories matching 'Jest' found. Current memories:"
# (lists all stored memories)
```
## Example 7: Multi-Project Memory Isolation
Each project maintains its own independent memory file. Memories from one project
do not leak into another.
```
~/project-a/.claude/memories/project_memory.json
→ "Use React + Next.js"
→ "Deploy to Vercel"
~/project-b/.claude/memories/project_memory.json
→ "Use Vue + Nuxt"
→ "Deploy to Cloudflare Workers"
When working in project-a:
→ Loads only project-a memories
→ Creates React/Next.js code
→ Deploys to Vercel
When switching to project-b:
→ Loads only project-b memories
→ Creates Vue/Nuxt code
→ Deploys to Cloudflare Workers
```
---
*[Tons of Skills](https://tonsofskills.com) by [Intent Solutions](https://intentsolutions.io) | [jeremylongshore.com](https://jeremylongshore.com)*
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |