groundy
articlessearch

agents & frameworks

117 articles·rss

Independent comparisons of agent stacks and multi-agent designs, tracking the gap between framework marketing and the failure modes that show up under real workloads.

latest
A partly open copper gate admits an ivory puzzle piece toward an interlocking ivory, gray and green sequence. Loose ceramic pieces wait beside it on textured ivory paper.

Do More Tools Make LLM Agents Worse? When to Gate External Evidence

Preprints show forced evidence gathering raised clinical agent error from 28.3% to 34.3%, while curated tool menus lifted ToolBench success to 0.898, suggesting gating is key.

9 min
11 min

How Cloudflare Detects MCP Traffic and What Enterprises Should Filter

12 min

How Spotify Cut Claude Code Token Usage 90%: What It Takes to Replicate

11 min

LangGraph vs CrewAI vs AutoGen: Which Python Agent Framework to Pick

archive1–50 of 113 · page 1 / 3
  1. Claude Code Auto Mode Is Broken: What to Gate Before Running Opus 5 Unattended
  2. Why Deep Research Agents Abandon the Plan Mid-Search
  3. MCP vs Function Calling: Who Should Own Your Agent's Tool Layer?
  4. Qwen-Agent Stretches 8k to 1M Context: Do You Need a Long-Context Model?
  5. Can AI Agents Do Root Cause Analysis? What Cloud-OpsBench Measures
  6. Which AI Agents Behave Badly: Cloudflare's Agentic Internet Data
  7. Compressing LLM Agent History: Pixel Rendering vs Summarization
  8. Anthropic A/B Tests Claude Code Effort Levels: Your Agent Is the Control Group
  9. Self-Hosted Coding Agents: The Safety Burden You Inherit
  10. Claude Code Weekly Limits Promo Ends in August 2026: Budgeting for Agent Teams
  11. Vibe Coding vs Control: How Developers Actually Used AI Coding Agents
  12. Multi-Agent or Single-Agent LLM: What Skill Distillation Actually Costs
  13. Multi-Agent LLM Systems Drift Into Misaligned Communication Over Long Horizons
  14. Sizing Agent Memory: A Capacity Planning Rubric for Long-Horizon LLMs
  15. When Spec-First Agents Dismantle Invariants: A Governance Case Study
  16. Cloudflare Kitesurf: V8 Isolates vs Containers for Agent Browsing
  17. Claude Code Skills vs Model Weights: Where Should Agent Skills Live?
  18. Cloudflare Precursor: Behavioral Detection for AI Agents
  19. Why Multi-Agent LLM Delegation Concentrates Risk
  20. Harness vs Scaffold: Why Claude Code and LangGraph Are Not Interchangeable
  21. MCP Tool Discovery Moves From Hardcoded Config to Runtime Agent Search
  22. Code-as-Action Agents Beat GAIA But Require Runtime Sandboxing
  23. Why Agent Security Tests Must Audit Full Trajectories, Not Single Turns
  24. x402 Per-Call Payments: Agent Wallet Custody and Replay Risks
  25. CodeRabbit Review Study: 56% Rejection Rate Demands Targeted Scoping
  26. LLM Agents Ignore Mid-Flight Halt Signals: 0 of 40 Trials Stopped
  27. Agent-First CLIs: Why GitHub, npm, and PyPI Must Publish Machine-Readable Contracts
  28. MCP and AGENTS.md Standardize Context, Not Agent Coordination
  29. Cursor's Swarm Math: When Cheap Agents Save Money and When They Fail
  30. Runtime monitoring beats alignment for agent-to-agent coercion
  31. Cloudflare's Agent Stack: Edge Trust, Identity, and Metering
  32. LM Studio Bionic vs Claude Code: Local-First vs Cloud Agent Tradeoffs
  33. Can a Malicious AGENTS.md File Compromise Your Coding Agent? A Threat Model
  34. How GitHub Projects Actually Adopt Coding Agents: New Empirical Data
  35. Why CLI Coding Agents Derail Mid-Run, Not at the First Mistake
  36. Test-Time Scaling Cost Falls as PRMs Reuse Generator KV-Cache
  37. Claude Code Skills vs Cursor Rules vs MCP: How Agent Skill Systems Compare
  38. TTHE: Test-Time Harness Evolution Changes the Test-Code Contract for Coding Agents
  39. Git-for-Data for Agentic Lakehouses: Why Agents Need Versioned State
  40. Game Theory Can Cut Multi-Agent LLM Hallucination, But Only If Payoffs Align
  41. WebSwarm: Recursive Multi-Agent Search vs Flat Orchestration
  42. Do Coding Agents Memorize Their Benchmarks? DeepSWE Tests on Unseen Tasks
  43. Can a 4B Model Run a Coding Agent? Terminus-4B vs Claude and GPT-4o
  44. Can You Prove a Governed AI Agent Actually Ran the Action You Authorized?
  45. AgentTether Repairs LLM Agent Failures with a Runtime Graph
  46. Can Multi-Agent LLM Negotiation Protocols Trust Their Own Samplers?
  47. Why Your AI Travel Agent Would Book a Bullfight
  48. Can Multi-Agent RAG Run Air-Gapped? A Forensics System Shows How
  49. Can You Prove an Agentic Trading Pipeline Has No Look-Ahead Bias?
  50. How Far Ahead Can a Coding Agent Plan? The Horizon Bottleneck
about this beateditorial framing

Agent frameworks ship faster than the rigor operators need to run them. Vendor docs promise orchestration, memory, and tool use; academic benchmarks and production post-mortems keep exposing the same structural gaps: diversity collapse in multi-agent ideation, hallucination amplification across consensus topologies, missing per-step rationale traces, role-based retry losing to graph-state failure isolation on long tasks, and configuration surfaces that punish static templates. This beat covers that delta.

The second through-line is governance and trust. Skill registries, tool-use protocols, and capability manifests are accumulating faster than auditable contracts for them. Trust schemas, contractual skill specs, and information-flow controls are arriving as bolt-ons rather than primitives, while the infrastructure layer — sandbox execution, private networking, agent memory — keeps absorbing functionality the framework layer used to own. The question of where the agent stack actually lives, and who is liable when it misbehaves, stays unresolved.

Coverage is comparative and opinionated. When a benchmark or paper exposes a gap that a major framework cannot close without redesign, that gets named. When a vendor ships governance theater rather than enforcement, that gets named too. The goal is help readers pick stacks that survive contact with production, not a taxonomy of every framework release.