Blog

Practical writing on AI coding guardrails — spec-driven development, agentic PR reviews, policy-as-code, observability, and product changelog updates from the Zenable team.

Don't see a feature you want? Let us know!

AllChangelogProEnterpriseFreeAI Native
ChangelogProEnterprise

How I learned to stop worrying and let my requirements self-improve

Agent Policies let provably good AI proposals through on their own: quality-scored against rubrics you own, bounded by budgets, and never skipping your Human-in-the-Loop approval workflows.

ChangelogProEnterprise

Human in the loop, but only when you want it

Approval workflows hold any matching change (from a teammate, a script, or an AI agent) until the humans your policy names have signed off.

ChangelogFreeProEnterprise

More agents, faster hooks, and a CLI that heals itself

Five new IDE integrations, one hook call per tool batch, sub-agent feedback triage, skill telemetry, signed Windows binaries, and a self-updating, self-healing CLI.

ChangelogProEnterprise

Agents + Custom Guardrails = Quality

Zenable learns your team's patterns and proposes them as requirements, now easier than ever to review, delegate, and approve on a fully redesigned Proposals page, from the UI, CLI, or MCP.

ChangelogProEnterprise

Your code audit shouldn't take three months

Interview-driven codebase assessments that produce a polished, evidence-backed report and leave your Zenable tenant bootstrapped with custom requirements, scopes, and guardrails. Coming soon; email hello@zenable.io for early access.

ChangelogProEnterprise

Self-Improving Guardrails

Your guardrails get sharper over time, on your terms: humans set the standards, the grading rubric, and the tolerances, while agents do the grading and propose refinements within them.

ChangelogProEnterprise

Custom roles for how your org actually works

Compose roles from 47 granular permissions (and growing!), define them per workspace or company-wide, and promote or demote between the two, with privilege containment enforced server-side.

ChangelogProEnterprise

Generate ESLint configs from your requirements

Zenable now generates ESLint configurations from plain-English requirements: deterministic, zero-latency enforcement in the lint step JS/TS teams already run.

ChangelogEnterprise

Sign in with Okta, and let your agents sign in too

Okta SSO for your team, any OIDC provider you like, and real logins for your AI agents and CI pipelines.

ChangelogProEnterpriseFree

If you're sick of hearing about speedups, sorry. Not sorry.

KV caching across every model that supports it, more bbolt tuning, and a parallelized AI review API. Three more wins for inference speed and CLI responsiveness.

ChangelogFreeProEnterprise

Full Windows support for the Zenable CLI

The Zenable CLI now runs natively on Windows -- same install flow, same hooks, same guardrails, no WSL detour required.

ChangelogProEnterprise

Roll out new requirements, one step at a time

Three enforcement modes -- audit, warn, and enforce -- so you can roll out new requirements incrementally without breaking your team's flow.

Discover AI Coding Guardrails
HomeWhy ZenableFeaturesHow it worksPricingContact UsBlogWorkshops
DocumentationTerms of ServicePrivacy PolicySecurity
Zenable 2026. All Right Reserved.