Agents are now very good at writing code. Agents are also composed of code (the harness). This means that agents can self-adapt through rewriting their harness - now supported in Letta Code through *mods*.
Mods are very similar to extensions in @badlogicgames's Pi harness. You
Agents can now not only rewrite their memory, but also rewrite the execution code they run in (the harness) through Mods.
Mods (inspired by Pi's extension system) allow for harness-level changes like:
- modifying context
- injecting custom tools
- customizing the statusline
Harness is memory.
My current running example: "please use uv not pip".
This can get written into system-level memory (context-as-memory), or if the agent has access to mutate its own harness, it can be written into a pre-tool use hook (block commands w/ pip).
Agents can now not only rewrite their memory, but also rewrite the execution code they run in (the harness) through Mods.
Mods (inspired by Pi's extension system) allow for harness-level changes like:
- modifying context
- injecting custom tools
- customizing the statusline
Agents can now not only rewrite their memory, but also rewrite the execution code they run in (the harness) through Mods.
Mods (inspired by Pi's extension system) allow for harness-level changes like:
- modifying context
- injecting custom tools
- customizing the statusline
External mods can also be installed through npm or git with: `letta install <source>`
We're listing community and official mods at letta.com/agent/mods/
Very exciting times in research for token-space continual learning, with ideas like dreaming finally hitting the mainstream.
One promising research direction we're excited to share more on is memory models, models that are trained explicitly for sleep-time compute / dreaming.
Letta's office hours livestream is today at 11:30am Pacific.
Join our Discord to hear about the latest in Letta, including:
- The mod registry
- Our dedicated GLM 5.2 endpoint
- Signal support
- The Agent SDK
We've been busy this week.
Discord link below 👇
Introducing letta-mcp-mod, a mod for @Letta_AI ‘s Letta Code to bring MCP support without bloating your context.
Letta Code doesn’t natively support MCP. That’s cool, and I appreciate their principled stance. Skills + bash work well. But there are some use cases for MCP, too.
New Letta Office Hours is up.
This week: @just_cameron is back, Letta Code harness mods, Windows fixes, Slack channel controls, Signal progress, local mode, and schedules improvements.
This clip covers the update segment before our colleague Caren joins for the mods demo + the
Come join our Office Hours livestream at 11:30am Pacific!
@just_cameron is back and ready to show off some of the amazing work the Letta team has done over the past few weeks.
Join us on Discord, link below 👇
I used the tutor and now all I do is ask Letta to do stuff all day. Night and day difference from codex because Letta *understands my culture by having not only long term memory, but an excellent way to update it*.
It's like having continual learning - check it out:
You can now create a "Tutor" agent in Letta.
Tutor walks you through how Letta works while learning your needs and tracking your progress. It proactively demos features - and explains how they work as it goes.
You can also ask Tutor to configure channels, schedules, skills, and
You can now create a "Tutor" agent in Letta.
Tutor walks you through how Letta works while learning your needs and tracking your progress. It proactively demos features - and explains how they work as it goes.
You can also ask Tutor to configure channels, schedules, skills, and