dep-radar
Load to run or tune a dependency sweep.
vanillagreencom · AI review on file
Installation
Read the source and dependencies before running a command. Copying does not install or execute anything.
npx skills add vanillagreencom/kendex --skill dep-radarAbout this skill
name: dep-radar description: "Load to run or tune a dependency sweep." summary: "Sweeps every pinned version in the repo (deps, SDKs, vendored forks, model weights), checks upstream, and lands upgrades with their fallout in one PR per surface." license: MIT user-invocable: true dependencies: required: [github] optional: [worktree] metadata: author: vanillagreen source: kendex repository: "https://github.com/vanillagreencom/kendex" bugs: "https://github.com/vanillagreencom/kendex/issues" version: "1.0.0" tags: [release]
Project Instructions
Problems with a kendex-owned skill go through kendex report; check ownership in the file first.
dep-radar
inventory → detect → research → classify → upgrade-with-fixes → report the owner tier.
Generic engine only: concrete package names, pinned binaries, and fork lists live in the per-repo inventory, never here.
Load github before Phase 4 (PR creation, CI status, merges). Load worktree when a run applies more than one surface, one working copy per surface branch.
Operating policy (the contract with the product owner)
Rule keys are the contract; an inventory owner-rule cites them. Dropping or renaming a key changes the contract.
| Rule | Contract |
|---|---|
auto-with-fixes | The default. Security fixes; patch/minor bumps; pinned-binary version+SHA refreshes from OFFICIAL manifests only; SDK, agent-tooling, and runtime-binary bumps and npm/cargo majors, doing the bump AND fixing its fallout (API migrations, re-vendored bundled-extension bridges, tests, CI) in the SAME per-surface workstream; bundled-extension fork updates and local patch rebases when the consuming repo's full test suite gates the sync. |
report-never-auto | Model-weight swaps; changes to durable/recorded data scope; anything an inventory owner-rule explicitly demotes. Nothing else is report-by-default. |
uncertain | Attempt the upgrade; report only what actually failed, with error output. |
defer | Only on a strong concrete blocker you actually hit, such as a dropped capability with no migration path or a transitive that does not support the new version. Never a generic "it's a major" risk or any other anticipated one. |
one-pr-per-surface | One PR per surface; never batch surfaces. A surface's fallout fixes go in THAT surface's PR. |
upstream-check-required | Every pinned surface must have a wired upstream check command; a surface lacking one is an inventory defect the run must fix. |
dated-report | Every run ends with a dated report. |
demote-only | Inventory owner-rules may demote auto→report, never promote report→auto. |
Phase 0: inventory (self-maintaining)
docs/dep-radar/inventory.md carries one row per pinned surface: pin location, upstream check command, refresh procedure, verify command, risk tier, applicable playbook, and any repo-specific owner rules.
First run (no inventory): sweep the repo for pins in package manifests and lockfiles, vendor/ dirs, SHA-256 constants near download or pin code, model manifest scripts, and version constants referencing upstream releases. Write the inventory, wire an upstream check per surface, and have the owner review the tiers.
Every run: diff discovered pins against the inventory, add new surfaces (each with a check), drop removed ones, and note the change in the run report.
Phase 1: detect
Read docs/dep-radar/last-seen.json (create if absent) and run each surface's upstream check. If nothing moved since last-seen, update checked_at, write a one-line report, and stop.
Phase 2: research
For each changed surface, read the changelog or release notes, never infer from version numbers. Extract breaking changes, deprecations, security fixes, new capabilities, and anything touching a contract the inventory names for that surface.
Phase 3: classify
Sort every finding per the operating policy plus the inventory's per-surface tier and owner rules.
Phase 4: apply the auto tier
Apply the inventory's refresh procedure, then fix the fallout in that surface's PR: migrate changed APIs, re-vendor bundled-extension bridges, repair broken tests and CI. Run the verify command; open the PR only once it passes locally, following the repo's review and merge-queue conventions. PR body: old→new version, changelog summary with links, fallout fixed, what was verified.
A blocker mid-apply or a failed verification stops the surface and makes it a report item with the exact error output, never ship a partial bump.
Phase 5: report
Write docs/dep-radar/report-<YYYY-MM-DD>.md, committed with the last-seen update: auto-applied bumps with PR links; blocked bumps with exact error output; the owner-decision tier; new capabilities unlocked. Each awaiting-decision item names the capability, what it unlocks, estimated effort and risk, and a recommendation. Surface the report to the owner (PR description or handoff doc), not just the file.
Technology playbooks
The inventory records which apply and every concrete package, binary, and fork name.
| Surface | Upstream check | Tier and handling | Verify |
|---|---|---|---|
| Pinned AI/agent SDK | Registry latest + release notes | Auto-with-fixes, majors included: migrate changed auth, runtime, and tooling APIs in the same PR. New provider models a bump exposes are report-tier; the bump itself ships. | Build + test suites; confirm expected models and features appear |
| Pinned runtime binary with SHA constants | Official release manifest for the exact version, never a third party, never hand-computed from a local download alone | Auto-with-fixes: migrate auth, protocol, and contract changes | Pin unit tests + a live download smoke on the host platform |
| npm/pnpm deps | pnpm -r outdated, pnpm audit | Auto-with-fixes, including major |