latent
agents· 8 min read
When submissions give orders
A conference submission talked an assistant into rewriting its own output. Testing two input gates exposed the tradeoff between catching attacks and rejecting legitimate work.
Ideas, experiments, and noteson data and AI.
A technical blog about AI agents, retrieval systems, databases, and geospatial tools. Explore how they work through code, experiments, and the decisions behind them.
Latest from latent space
View all
agents· 6 min read
php2shell: Are We Affected?
The php2shell vulnerabilities turn a PHP security alert into an urgent operational question. I built CVE Central to inspect deployed applications, confirm CVE applicability, and prepare a remediation plan.
embeddings· 7 min read
When meaning misses
A World Cup project vanished from search after one missing letter. The scores explain why embeddings, correct prompts, and hybrid retrieval still need a test of relevance.
Agents
Working agents for infrastructure automation, database-backed AI, and search evaluation.
Agent
Hermes homelab agent
A WhatsApp request, a reviewed plan, and a Docker-ready Proxmox container. Watch the build and an independent check of the result.
Hermes Agent · WhatsApp · Proxmox · Docker · Technitium
Explore the agent →
Agent
Mexico City transit assistant
A conversational transit map that resolves places and asks a routing engine to calculate the journey.
Gemini · MCP · PostGIS · pgRouting
Explore the agent →
Agent
JobMap
Describe the work you want to do, explore matching job postings, and inspect the evidence behind Jev's relevance judgments.
Jev · Sentence embeddings · PostgreSQL · pgvector
Explore the agent →