kenpusney/greathangpt-classical-chinese
Viewer ā¢ Updated ā¢ 3.6M ā¢ 50
Kimmy
kenpusney
Ā·
AI & ML interests
None yet
Recent Activity
updated a dataset 4 days ago
kenpusney/greathangpt-classical-chinese published a dataset 5 days ago
kenpusney/greathangpt-classical-chinese liked a Space 14 days ago
Xenova/the-tokenizer-playgroundOrganizations
None yet
updated a
dataset 4 days ago
published a
dataset 5 days ago
liked a
Space 14 days ago
reacted to aufklarer's post with š„ 5 months ago
Post
3465
Context Engineering for Code Agents: Why They Fail and How to Fix Them
Code agents don't fail because they can't code ā they fail because their context turns into a junk drawer.
I wrote a practical survey covering the emerging discipline of context engineering for agentic hybrid applications: the techniques, papers, and architectural patterns that keep long-running code agents on track as their token windows fill up with tool logs, stale diffs, and repeated file dumps.
What's covered:
Why long context windows alone don't save you (position bias, distractor sensitivity)
Observation masking vs. LLM summarization ā and when simple beats clever
Tool-output compression with approaches like LLMLingua-2
Trajectory reduction: pruning dead branches from agent history
Memory hierarchies: session ā working set ā notes ā cross-session
How MCP and standardized tool interfaces reduce context debt
Dynamic context policies trained with RL (DeepMiner, MEM1)
Meta-agent CI loops for measuring regressions across agent configs
The core argument: the engineering challenge isn't "make the model smarter" ā it's make the agent's context and verification smarter. That's where the real leverage is in 2026.
š Read the full post: https://blog.ivan.digital/context-engineering-for-agentic-hybrid-applications-why-code-agents-fail-and-how-to-fix-them-076cab699262
Code agents don't fail because they can't code ā they fail because their context turns into a junk drawer.
I wrote a practical survey covering the emerging discipline of context engineering for agentic hybrid applications: the techniques, papers, and architectural patterns that keep long-running code agents on track as their token windows fill up with tool logs, stale diffs, and repeated file dumps.
What's covered:
Why long context windows alone don't save you (position bias, distractor sensitivity)
Observation masking vs. LLM summarization ā and when simple beats clever
Tool-output compression with approaches like LLMLingua-2
Trajectory reduction: pruning dead branches from agent history
Memory hierarchies: session ā working set ā notes ā cross-session
How MCP and standardized tool interfaces reduce context debt
Dynamic context policies trained with RL (DeepMiner, MEM1)
Meta-agent CI loops for measuring regressions across agent configs
The core argument: the engineering challenge isn't "make the model smarter" ā it's make the agent's context and verification smarter. That's where the real leverage is in 2026.
š Read the full post: https://blog.ivan.digital/context-engineering-for-agentic-hybrid-applications-why-code-agents-fail-and-how-to-fix-them-076cab699262
- 2 replies
liked 2
models 6 months ago
liked a
dataset 6 months ago
liked 2
models 6 months ago
upvoted a paper 6 months ago
liked 5
models about 1 year ago
nomic-ai/colnomic-embed-multimodal-3b
Visual Document Retrieval ā¢ Updated ā¢ 11.8k ā¢ 39
nomic-ai/colnomic-embed-multimodal-7b
Visual Document Retrieval ā¢ Updated ā¢ 23.2k ā¢ 107
nomic-ai/nomic-embed-multimodal-7b
Visual Document Retrieval ā¢ Updated ā¢ 1.51k ā¢ 49
mrcuddle/live2d-model-maker
Text-to-Image ā¢ Updated ā¢ 101 ā¢ ā¢ 30
nari-labs/Dia-1.6B
Text-to-Speech ā¢ 2B ā¢ Updated ā¢ 5.53k ā¢ ā¢ 2.88k
liked a
model over 1 year ago
updated a
model over 1 year ago
published a
model over 1 year ago
liked a
model over 1 year ago
updated a
model over 1 year ago