for agents/llms.txtv0.13.110

Home / Articles / A policy for every skill / Versions / v1.3.0

A policy for every skill: what changed in v1.3.0

From v1.2.0 (2026-10-14, e1d5ff4a7) to v1.3.0 (2026-10-27, db67c5ce1), paragraph by paragraph.

0 paragraphs added, 0 removed, 2 changed in place, 46 unchanged. About 0 words added and 69 removed. Insertions are marked like this, deletions like this; unchanged runs are folded to one line; figures appear as their file names.

← v1.2.0 · all versions

2 unchanged paragraphs

!shot sk-pipeline.webp | images/ | A policy for every skill.skill: The natural-languagethe control at the top isand the canonical source;checks each contributed skill passes deterministic checks, a model that reads it in its context, rules an agent generated from the control, and a human review, before it reaches the organisation's repository and the developers who pull it.passes. | architecture

19 unchanged paragraphs, under In short {#in-short}, What Sean built {#built}, Skills are the new dependency {#dependency}…

!shot sk-context.webp | images/ | Same check, different context, different verdict. Two illustrative skills described in the columns of an agent behaviour policy: one that turns internal documentation into a post,policy, with trustedthe inputssame check and adifferent person pressing publish; one that researches the web, with untrusted inputs and a barrier that must treat fetched text as data.verdicts. | comparison

25 unchanged paragraphs, under Hope-based security, and why it is still worth writing {#hope}, Use models to need them less {#codify}, A catalogue is a graph {#graph}…

← v1.2.0 · all versions