Adding my voice to this thread, though from a slightly different angle than most. I use Recall as a , not a second one. My primary loop is podcast notes, YouTube takeaways, and curated web content flowing Recall as a context store for the topics I care about, and then flowing via MCP into Claude (and other agents) whenever I need that context. For me that means write access is genuinely not the top priority — I don't have a constant stream of new synthesis that needs to push back into Recall.

That said, there's one narrow write case that resonates with where Carl Webb is going, just at a smaller scale.

When MCP-Read pulls a specific Recall card into an LLM session and that card actually helps me reach an insight, I'd love to post a tiny feedback marker back to that card — something like Not a full synthesis note, just a lightweight breadcrumb.

The value isn't really for the current session. It's for the MCP read from a sister agent or sister project. The next agent pulling the same card would know it has already been used elsewhere, and what insight it produced. Over time that becomes a Single-Source-of-Truth signal — distinguishing original sources from content that has been recycled three loops around. A small protection against the failure mode of "I keep reading this in agent contexts, therefore it must be true." For my own use, even a minimal POST-a-tag or POST-a-comment endpoint would cover most of what I'd want.

I really love Recall and want to use it as my personal knowledge management memory store and foundation for my "second brain." I need to be able to route things into it FROM the rest of my life (email, Slack, Granola, Zoom recordings, voice notes, Zapier) INTO recall, with each incoming event triggering the autotagging and related processes. Apple "Share" can work if available, but allowing easier accessess for adding content would be a huge plus for me.

Write access is the key to turning Recall into a personal knowledge system. Not interested in CLI/MCP (can open source that myself), just the API. Would open up so many use cases.

This is a must have. I've imported a lot of data from other sources but stopped the process because it is cumbersome to go through the cards, tag and reorganizing them. That is all still a manual process. It has also kept me from committing to the Max plan as well as bringing on family and employees to use the tool collaboratively.

The core gap is that there's no way to programmatically add content to Recall from outside the browser. The knowledge base is only as powerful as what's in it, but ingestion is currently limited to whatever the browser extension can reach.

There's a whole category of content with no browser surface: media players, podcast apps, e-readers, CLI tools. For example, I want movies and TV shows I finish on Plex to automatically land in Recall as cards. The architecture is trivial (webhook to script to TMDB URL to Recall), but it's completely blocked without a write endpoint.

I'll be honest: I subscribed to Plus assuming write access was part of the deal. A read-only API and MCP is a strange offering for a paid tier. Pulling content out is useful, but pushing content in is the whole point.

Even a single POST endpoint that accepts a URL and triggers the same save-and-summarize flow as the browser extension would unlock a huge range of automation. MCP write support would be equally valuable for agent workflows.

Hoping this moves fast.

We need write access

Would love a cli my agent could interact with even more than an api but an API would work. I'm not a big fan of MCP these days but an api is fine. I'll build my own cli then ;-)

Really pleased to see read access land. Just wanted to add my voice to the write access request with a specific use case that I think sits beyond pure capture.

I use Recall alongside Claude (via MCP) as a thinking and synthesis layer. The read integration is already changing how I work — I can ask Claude to check Recall for prior context before starting new work, which surfaces connections I’d otherwise miss.

The gap right now is the loop back. A lot of my most valuable thinking happens inside Claude conversations — iterative reasoning, synthesising across sources, reaching a conclusion. Without write access, the output of that work either gets copy-pasted manually into Recall (friction that breaks the habit) or lives in chat history that becomes increasingly hard to retrieve over time.

Write access would close the loop: read context in, think it through, write the synthesis back as a new note with appropriate tags. Recall becomes a genuine second brain rather than a one-way archive.

Thanks for releasing the API & MCP with rebrand! But its also very important to have write ability. I need to be able to programmatically pipe saved youtube videos and IG saved posts into my recall account, and write access is required for that.