On-screen text
UNLIMITED
AI MEMORY
Tencent
TencentDB
Agent Memory
Agents remember. Humans innovate.
Migrating data from an older version
UNLIMITED
AI MEMORY
memory
Claude Code
any AI agent
Tired of repeating yourself? Tell Claude to
remember what you've told it using
CLAUDE.md.
Prefer the Terminal experience? Switch back in Settings.
.neumorphic-button:after {
content: ";
@apply absolute inset-0 opacity-0 transition-opacity duration-300 bg-gradient-to-br
[@968f7f]/20 to-transparent rounded-full;
}
.neumorphic-button:hover:after {
@apply opacity-100;
}
...
Claude Code v2.1.160
Opus 4.8 (1M context) with medium effort - Claude Pro
/Users/josephodoro
/fast
/focus
/framer
/feedback
/figma-use
/firecrawl
/framer-audit
/firecrawl-map
Toggle fast mode (Opus
Toggle focus view (show
summary, and the final
Use when the user wants to
publish a website or app
Submit feedback, report
conversation
(figma) MANDATORY: perform
(figma) invoke this skill BEFORE
Search, scrape, and import
Firecrawl CLI. Use this to
Run a batch of audits for
health, CMS integrity, SEO.
Discover and list all
Claude Website
Claude Website
When chat message
received
Get Memories
Search Record
Aggregate
Merge
combine
Editor
Executions
Agent
Style
New Chat
Ask Framer...
Claude Code v2.1.160
Tips for getting started
Run/limit to create a CLAUDE
No recent activity
Recent activity
- Let me read the most relevant articles to answer this.
- Read 10 files (ctrl+o to expand)
- Your biggest inspiration, based on what the wiki documents, is Naruto -- and it's
in terms of depth and duration of influence.
> bypass Permissions on (shift+tab to Cycle) - esc to Interrupt
Good afternoon, team vm
How do I manage all my documentation?
Sonnet 4.6
UNLIMITED
AI MEMORY
Close it
a new one
no idea
you are
prefer
you were
last week
So Tencent
tier
system
completely
of how
works
Analyze this document
You uploaded a document
Q2 Risk Assessment Report.pdf
PDF 1.2 MB
Now summarize only the risks
AI Assistant
Here are the key risks identified in the document:
Market volatility could impact revenue projections.
Supply chain disruptions may delay critical deliveries.
Regulatory changes may increase compliance costs.
Cybersecurity threats pose a risk to sensitive data.
Talent retention challenges could affect key initiatives.
memory
happening
Long term
stores
preferences
personality
over time
works
It watches
conversations
pulls out
like
prefers
Python or
wants
responses
and stores
that get
the longer
the agent
token
by 61%
task
because
Rereading
background
session
source and
locally
Memory
Capability
Benchmark
OpenClaw
Success
With
Plugin
Relativ
Δ
Short-term
WideSearch
33%
50%
+51.52%
Short-term
SWE-bench
58.4%
64.2%
+9.93%
Short-term
AA-LCR
44.0%
47.5%
+7.95%
Long-term
PersonaMem
48%
76%
+59%
These results are measured over continuous long-horizon
sessions, not isolated turns. For example, SWE-bench runs
50 consecutive tasks per session to simulate the context-
accumulation pressure of real-world long-horizon agents.
Overview
