Apr 16 · 11 min read · Verification Is Not Causal: Why Shared Context Erases the Admissibility Gap Maksim Barziankou (MxBv) PETRONUS™ | research@petronus.eu DOI: 10.5281/zenodo.19609707 Axiomatic Core (NC2.5 v2.1): DOI 10.17605/OSF.IO/NHTC5 When someone asks me what Conte...
LLaura and 2 more commented
Apr 8 · 7 min read · Most agent tooling mistakes coordination for reliability. It gives you more roles, more agents, more orchestration, and more shell theater. The demo gets more impressive. The system does not necessarily get easier to trust. That tradeoff used to be t...
Join discussionMar 23 · 5 min read · Last week, our pipeline produced a proofpack with decision: HUMAN_REVIEW. The contract had 10 acceptance criteria. The engineer agent created all the new files, build passed, tests passed, three independent reviewers ran. Everything looked correct — ...
Join discussionMar 18 · 5 min read · You can review an AI diff with three models and still have zero ground truth. They'll tell you what looks "reasonable", not what's correct. The failure mode isn't "bad code". It's unfalsifiable intent: the requirement never became something you can r...
Join discussionMar 17 · 6 min read · The morning digest surfaced Leanstral -- Mistral's open-source agent for formal verification in Lean 4. A mixture-of-experts model (119B total, 6.5B active per token) that scores within 80% of Claude Opus on the FLTEval theorem-proving benchmark at a...
Join discussionMar 15 · 6 min read · The first version of Signum ran in a single pass: CONTRACT → EXECUTE → AUDIT → PACK. If the audit found a problem — block. Human deals with it. An honest process, but a limited one. Imagine code review where the reviewer can only comment and the auth...
Join discussionMar 13 · 3 min read · OpenClaw Reached 68,000 GitHub Stars — Here's the Missing Piece OpenClaw just passed React. 68,000 stars. It's the most popular open-source browser automation framework for AI agents. But enterprise teams are asking a question the documentation doesn...
Join discussionMar 10 · 5 min read · You asked an AI research assistant a detailed question and got a confident multi-page answer with citations. Some of those citations don't exist. Several facts contradict each other. The synthesis reads well — it's structured, well-argued, fluent. It...
Join discussionMar 6 · 6 min read · AI made writing code mass-affordable. It did not make trusting code any cheaper. The standard pipeline today: issue or prompt, AI writes code, AI or human reviews the PR, merge. This was always imperfect, but it scaled when humans wrote every line an...
Join discussion
