Skip to content

AI readiness toolkit

v1.8.0 Feature

This release adds 3 notable features for engineering teams evaluating rollout.

✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Summary

AI summary

Added a new Layer 1 Runtime Legibility/Liveness check and enriched Layer 0 graph‑based navigability to the /assess readiness model.

Changes in this release

Feature Medium

Adds new Layer 1 Runtime Legibility / Liveness signals to /assess.

Adds new Layer 1 Runtime Legibility / Liveness signals to /assess.

Source: llm_adapter@2026-05-28

Confidence: high

Feature Medium

Enriches Layer 0 with graph‑based navigability and centrality×staleness signals.

Enriches Layer 0 with graph‑based navigability and centrality×staleness signals.

Source: llm_adapter@2026-05-28

Confidence: high

Feature Medium

Adds second heatmap showing documentation staleness.

Adds second heatmap showing documentation staleness.

Source: llm_adapter@2026-05-28

Confidence: high

Feature Medium

Reframes scoring from /7 to /8 across three dependency‑ordered bands.

Reframes scoring from /7 to /8 across three dependency‑ordered bands.

Source: llm_adapter@2026-05-28

Confidence: high

Bugfix Medium

Fixes huddle to send team‑mode messages per‑recipient instead of broadcasting.

Fixes huddle to send team‑mode messages per‑recipient instead of broadcasting.

Source: llm_adapter@2026-05-28

Confidence: high

Bugfix Medium

Resolves latent plugin‑path bug; skills now resolve bundled scripts via $CLAUDE_PLUGIN_ROOT.

Resolves latent plugin‑path bug; skills now resolve bundled scripts via $CLAUDE_PLUGIN_ROOT.

Source: llm_adapter@2026-05-28

Confidence: high

Full changelog

/assess: the read-side foundation

Closes the largest gap in the /assess readiness model — the read side. The model now measures whether an agent can form a true picture before it acts, and reframes every layer as a truth-pressure check: scaffolding is scored on whether it is kept honest, not merely present.

Highlights

  • New Layer 1 — Runtime Legibility / Liveness. Intra-repo dead-code candidates plus a three-rung observability check (instrumented → discoverable → agent-reachable). Distinguishes "telemetry exists but the agent can't query it" from genuinely reachable runtime truth.
  • Enriched Layer 0 — graph-based navigability. A doc link-graph (PageRank hubs, orphans, reachability from entry points, MOC validation, doc→code edges) and a centrality×staleness signal that surfaces the most dangerous "lying map" — a load-bearing doc that has frozen while its subject code churns.
  • Second heatmap — docs staleness. Inverted encoding of the complexity treemap: size = link-graph centrality, hue = doc staleness, saturation = churn of the code the doc describes. Vivid red = a frozen doc beside churning code.
  • Reframed scoring /7/8 across three dependency-ordered bands (read-side foundation → write-side enforcement → meta), with rescaled maturity bands. Report shape stays backward-compatible.
  • Also resolves a latent plugin-path bug: skills now resolve bundled scripts via $CLAUDE_PLUGIN_ROOT, fixing clean plugin installs.

Both the Claude Code plugin and the standalone Desktop/Cowork ZIPs are updated (the standalone ZIPs are on the standalone-skills-latest rolling release).

What's Changed

Features

  • feat: Add truth-pressure read-side signals to /assess (navigability + liveness, /8 model) by @bjcoombs in https://github.com/bjcoombs/ai-native-toolkit/pull/32

Fixes

  • fix(huddle): send team-mode messages per-recipient, not broadcast by @bjcoombs in https://github.com/bjcoombs/ai-native-toolkit/pull/31

Full Changelog: https://github.com/bjcoombs/ai-native-toolkit/compare/v1.7.1...v1.8.0

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Track AI readiness toolkit

Get notified when new releases ship.

Sign up free

About AI readiness toolkit

All releases →

Related context

Beta — feedback welcome: [email protected]