Skip to content
Tagged articles

tool contracts

4 articles · Page 1 of 1
Ray's Galactic Tech
Ray's Galactic Tech
Oct 22, 2026 · Artificial Intelligence

AI SRE Agents: Correct Answers ≠ Safe Execution — Layered Evaluation & Release Gates

This article presents a comprehensive evaluation framework for AI SRE agents that audits tool usage, execution trajectories, evidence provenance, safety constraints, and environmental outcomes — not just final answer correctness — using deterministic contract tests, real-model trials, mutation testing, and statistical release gates to ensure reliable, auditable automation.

AI SRE AgentCI/CDGo
0 likes · 62 min read
AI SRE Agents: Correct Answers ≠ Safe Execution — Layered Evaluation & Release Gates
Architect
Architect
Oct 8, 2026 · Artificial Intelligence

Beyond the Model: Dissecting the Harness Layer of 11 Coding Agents

This article analyzes the harness layer of 11 coding agents—including Claude Code and Codex—based on an 83-page arXiv paper, revealing how runtime systems manage tools, context, permissions, and verification beyond the model itself.

Claude CodeCodexHarness Engineering
0 likes · 32 min read
Beyond the Model: Dissecting the Harness Layer of 11 Coding Agents
Architect
Architect
Aug 10, 2026 · Artificial Intelligence

Anthropic Deep Dive: Context Engineering Lessons from Real‑World R&D

The article analyzes Anthropic’s “Effective context engineering for AI agents,” showing how larger context windows can degrade, categorizing information by stability, designing prompts in the Goldilocks zone, structuring tool contracts, and applying runtime information scheduling, compression, structured notes, and sub‑agents to keep AI agents reliable in complex development workflows.

AI agentsAnthropicLLM
0 likes · 19 min read
Anthropic Deep Dive: Context Engineering Lessons from Real‑World R&D
Shuge Unlimited
Shuge Unlimited
Jun 18, 2026 · Artificial Intelligence

What the 120k‑Character Claude Fable 5 Prompt Leak Reveals About Its True Architecture

A leaked 120 KB system prompt for Claude Fable 5 shows that the model is not merely a chat bot but a fully engineered agent system with layered responsibilities, tool contracts, hard and soft constraints, runtime patches, and an opt‑in design that prevents it from autonomously selecting commercial partners.

Claude Fable 5LLM constraintsMCP
0 likes · 17 min read
What the 120k‑Character Claude Fable 5 Prompt Leak Reveals About Its True Architecture