# Skill: Save Memory

The **save_memory** skill teaches Claude or Codex how to store content from the current conversation into Context Link as a **Memory**. Memories are living documents saved under any `/slash` route — like `/brand-voice`, `/roadmap`, or `/meeting-notes` — that any AI session can retrieve and build on later.

Instead of losing a great output when you close the chat, you tell your AI to save it. Next time — in a new conversation, a different project, or even a different AI tool — you (or your AI) can fetch it back.

## What are Memories?

Memories are the **WRITE** side of Context Link. While the [get_context skill](/content/docs/skill-get-context/index.html) handles READ (searching your connected sources), Memories give your AI a separate workspace to save and maintain its own documents.

Key things to know:

- **Saved under any `/slash` route** — `/brand-voice`, `/keyword-tracker`, `/support-faq`, `/roadmap`, whatever makes sense
- **Separate from your synced sources** — Memories never modify your Notion pages, Google Docs, or websites. They live alongside them but are completely independent.
- **Shared across sessions** — save something in Claude, retrieve it tomorrow in ChatGPT, Copilot, or any other AI tool that can reach your Context Link
- **Versioned** — saving to the same slug creates a new version. The latest always wins on retrieval, but old versions are preserved.

## When to use this skill

- You've had a productive conversation and want to keep the key takeaways for future sessions
- You want to build up reusable assets over time — brand voice docs, style guides, FAQ answers, project specs
- You're summarising research and want it available for future AI sessions across any tool
- You want to save something under a specific topic name so it's easy to find later

For updating content that's already been saved, see the [Update Memory skill](/content/docs/skill-update-memory/index.html).

## How it works

When triggered, the skill:

1. Picks a slug based on what you asked — or uses `chat-session-{datetime}` for generic "save this conversation" requests, or infers one from the conversation topic
2. Distils the conversation into a clean, reusable markdown document (stripping conversational back-and-forth, keeping decisions, specs, and key details)
3. POSTs it to your Context Link under that slug

The content is then chunked, embedded, and available for retrieval immediately — via the get_context skill, a direct link, the ChatGPT connector, or the API.

## The skill file

Below is the full skill definition. Replace `YOUR_CONTEXT_LINK` with your personal Context Link URL (e.g. `yourname.context-link.ai`). If you use a PIN, append `?p=YOUR_PIN` to the URL.

````markdown
**What is Context Link?** Context Link is an external service that indexes connected sources (websites, Google Drive, Notion) and memories into a searchable knowledge base. It provides semantic search and memory storage via a simple URL: `subdomain.context-link.ai/query?p=optional_pincode`. If you don't know the user's Context Link URL, ask them for it.

---

### Save to Context Link

Save content from the current conversation to Context Link. One request, no fuss.

**Workflow:**

1. **Pick the slug.** If the user said "save {name}", use that name as the slug (lowercased, dashed). If the user said "save this conversation", "save this chat", or similar without a specific topic name, use `chat-session-{YYYY-MM-DD-HHMM}` as the slug. Otherwise, infer a short descriptive slug from the topic. Only ask if truly ambiguous.

2. **Summarize the content.** Distill the conversation into a concise, reusable reference document in markdown. Focus on what's useful to retrieve later — strip conversational back-and-forth, keep decisions, specs, and key details.

3. **Print this message:** `🔗 Saving memory to Context Link → {SLUG}` — Never print the actual Context Link URL, as it contains a private 'pin' or 'p' URL param.

4. **POST it.**

```bash
curl -s -X POST "`YOUR_CONTEXT_LINK/{SLUG}`" \
  -H "Content-Type: text/plain" \
  -d 'Your markdown content here'
```

The body is raw text/markdown — not JSON. The server's `LlmPoweredParser` handles chunking and structuring internally. Just send clean markdown.

The slug goes in the URL path. Replace `{SLUG}` with the chosen name (lowercase, use-dashes-for-spaces).

**Success response:** `{"message": "Saved", "namespace": "{SLUG}"}` with HTTP 201.

**Rules:**
- **Keep the body under 100KB.** If content is longer, summarize or condense it before sending. Do NOT split into multiple requests — distill into one concise document.
- If the user says "save X" — X is the slug. Always. No questions asked.
- Do NOT test the endpoint first. It works. Just POST.
- Do NOT verify by fetching it back. Trust the 201.
- Do NOT send multiple requests. One POST, done.
- Saving to the same slug creates a new version (latest wins on GET).
- After saving, confirm briefly: "Saved to Context Link as `{SLUG}`."
- If the request is blocked, ask the user to add `*.context-link.ai` to Claude's **Settings → Capabilities → Domain Allowlist** (or select "All domains"), then retry.
````

## Installation

### Claude (Chat, Cowork, and Code)

Skills work with **Claude Chat**, **Claude Cowork**, and **Claude Code**. (If you're using Claude Cowork, you can also use the [Context Link Plugin](/content/docs/claude-plugin/index.html) to install all skills at once.)

To install this skill:

1. In **Claude Chat** or **Claude Cowork**, click **Customize** in the top right
2. Click **Skills**, then **+**, then **Upload a skill**
3. Upload the skill file (download from [Installation](/content/integrations/index.html))
4. **Allow network access** (see below)

For **Claude Code**, place the skill file in your project's `.claude/skills/` directory.

#### Allow network access

For Claude to reach Context Link, you need to allow the domain in Claude's settings:

1. In Claude, go to **Settings → Capabilities → Domain Allowlist**
2. Add `*.context-link.ai` to the allowlist (or select "All domains")

Once installed, tell Claude to **"save this as \[topic\]"** and it handles the rest.

### OpenAI Codex

1. Download the skill files from [Installation](/content/integrations/index.html)
2. Unzip and place the `save_memory` folder into `~/.agents/skills/` (or your repo's `.agents/skills/` directory)
3. Codex auto-detects the skill

See the [Codex skills documentation](https://developers.openai.com/codex/skills/) for more on skill file locations and priority.

## Example usage

**Saving a brand voice doc after refining it in conversation:**

> **You:** Save this as brand-voice
>
> **AI:** _Summarises the brand voice guidelines from the conversation, POSTs them to your Context Link under `/brand-voice`._ "Saved to Context Link as `brand-voice`."

**Saving the whole conversation for later:**

> **You:** Save this conversation to Context Link.
>
> **AI:** _Uses `chat-session-2025-01-15-1430` as the slug, distils the conversation into clean markdown, and saves it._ "Saved to Context Link as `chat-session-2025-01-15-1430`."

**Saving research for later use across tools:**

> **You:** Save this to Context Link — I want to reference this competitor analysis tomorrow.
>
> **AI:** _Infers the slug `competitor-analysis`, distils the key findings into clean markdown, and saves it._ Now retrievable from any AI tool via your Context Link.

**Building up a knowledge base over time:**

> **You:** We just nailed down our Q2 roadmap priorities. Save this as q2-roadmap.
>
> **AI:** _Extracts the roadmap decisions, saves under `/q2-roadmap`._ Next week, any team member can ask their AI to "get context on q2-roadmap" and get the latest version. Or for a direct answer, ask Context Link a question outright — _"ask Context Link: what are our Q2 roadmap priorities?"_ — and get back one short paragraph with numbered citations. The [Ask Question skill](/content/docs/skill-ask-question/index.html) is a Pro feature and counts toward your monthly LLM allowance.

## Slugs and versioning

- Slugs are case-insensitive and normalised (spaces become dashes)
- Saving to the same slug creates a **new version** — the old content is preserved, but retrieval always returns the latest
- Use descriptive slugs: `/brand-voice`, `/q2-roadmap`, `/onboarding-checklist`, `/keyword-tracker`
- Memories are searchable alongside your synced sources — so when you "get context on brand voice", both your saved Memory and any relevant synced docs come back

* * *

Don't have a Context Link account yet? [Sign up and connect your first source](/content/site-root.html) — start saving and retrieving your AI's best outputs in minutes.