Home / Articles / Policy as code needs a policy first / Versions / v1.1.0
Policy as code needs a policy first: what changed in v1.1.0
From v1.0.0 (2026-10-14, 484d23815) to v1.1.0 (2026-10-27, db67c5ce1), paragraph by paragraph.
0 paragraphs added, 0 removed, 3 changed in place, 58 unchanged. About 0 words added and 108 removed. Insertions are marked like this, deletions like this; unchanged runs are folded to one line; figures appear as their file names.
2 unchanged paragraphs
!shot pq-altitudes.webp | images/ | One policy at three altitudes. At the top,altitudes: the business's words: a board policy, a row of a control spreadsheet, an article of a regulation. In the middle,words, the ABP row for one agent in one deployment: reach, mandate, gap, barrieragent, and owner. At the bottom, the code where decisions are made: a Rego rule, a hook, a validator check. Generating code walks down; reverse engineering walks up; a node with no edge is a gap.made. | architecture
23 unchanged paragraphs, under In short {#in-short}, Policy as code, and the question it answers {#question}, What an agent behaviour policy adds {#adds}…
!shot pq-session.webp | images/ | This session's own policy, reverse engineered from its code.code: Eleveneleven rules; for each,rules, the code thatbehind enforces it, who can change that code,each and the barrier that results. Eight have some code behind them; only one, never pushing to sgit.ai, has code the agent cannot change.barrier. | comparison
6 unchanged paragraphs, under Generating code from the policy {#generate}
!shot pq-roundtrip.webp | images/ | A round trip through Open Policy Agent.Agent: Fivefive ABP rows as data, generated into aRego, 35-line Rego policy, evaluated against seven requests,evaluated, then parsed back with OPA's own parser and compared with the ABP: one row with no code, one rule with no row.compared. | architecture
27 unchanged paragraphs, under From spreadsheets and paper {#paper}, Why it needs fractal semantic graphs {#fsg}, The four jobs {#jobs}…