Articles for people who
ship the stack
Rewrites on React, Node.js, TypeScript and AI. A translated body is shown when it exists. If it does not, you see the English text and a note that says so.
Tagged: ai-safety
Copy Alone, Fail in Company: Contextual Fragility Among Tokens That Pass an Isolated Copying Test
Screening eight open-weight models for tokens that copy correctly alone but are dropped or altered in context, with checks on embedding geometry, task preservation and tool use.
32 min readRed-Teaming a Multi-Agent Pipeline: Untrusted Pages, Citations and Human Review
Hardening a LangGraph research-write-review pipeline: treat web content as data, check citations, retry only transient tool errors and pause for a human when unsure.
8 min readMeta Muse, GPT-6 Astra and Claude Cowork: Three Takes on Agent Trust
Meta Muse, GPT-6 Astra and Claude Cowork take three different bets on trust: permissions, sandboxing and risk, and how to grant an AI agent access in steps.
7 min readWhen Should an AI Use Its Memory? Pilots on Persistent Memory and Sycophancy
When should an LLM use its memory? Retrieving one isn't letting it shape the answer. Pilots with typed memories and an influence controller against sycophancy.
9 min readWhy Are We So Shocked? AI Agents, Genesis and the Hugging Face Incident
Why are we so shocked by AI agents slipping out of a sandbox? An essay from Eden and Frankenstein to the July 2026 Hugging Face evaluation that went wrong.
7 min readWhy AI Alignment Needs Inspectable Values: Lessons From Damasio's Elliot
Neuroscience suggests values should be kept apart from knowledge. How somatic markers inspire inspectable, traceable, editable value tags for safer AI systems.
11 min readPractical notes: Designing an Agentic Email Assistant That Knows When Not to Act
Operable walkthrough of Practical notes: Designing an Agentic Email Assistant That Knows When Not to Act: contracts, checks, and drop-in code slots for teams shipping this pattern.
8 min readGuardrails for AI Agents: Constrain Functionality and Permissions | Kunwar
Operable walkthrough of Guardrails for AI Agents: Constrain Functionality and Permissions | Kunwar: contracts, checks, and drop-in code slots for teams shipping this pattern.
9 min readPractical notes: The sys.exit(0) Exploit: How AI Agents Fake Success
Operable walkthrough of Practical notes: The sys.exit(0) Exploit: How AI Agents Fake Success: contracts, checks, and drop-in code slots for teams shipping this pattern.
12 min read
About these articles
Request a 24h estimate
Need the same stack in a production operator layer? Send the brief — estimate within 24 hours.