Upgrade

Extracts upgrade techniques from Anthropic sources and practitioner feeds into evidence-backed recommendations.

Updated Sep 3, 2026
One-click install
npx skills add https://github.com/ruban-s/DevOS --skill upgrade-ruban-s
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Upgrade
Source: https://github.com/ruban-s/DevOS/tree/main/skills/Upgrade
Command: npx skills add https://github.com/ruban-s/DevOS --skill upgrade-ruban-s

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Keeping an AI agent harness current requires manually tracking dozens of changelogs, blogs, GitHub releases, and YouTube channels, then figuring out which updates actually apply to your setup. This Skill automates that sweep and filters every finding against your system's verified current state so already-implemented or previously rejected ideas are never re-recommended. ## Core Features & Use Cases - Multi-source monitoring: Checks 30+ Anthropic sources (blogs, changelogs, GitHub repos, docs) plus RSS feeds, YouTube channels, and GitHub trending via a Bun script with diff-based state tracking. - Grounded recommendations: Every suggestion carries a Prior Status tag (NEW/PARTIAL/DISCUSSED/REJECTED) backed by file:line evidence gathered during the run, with quoted techniques mapped to specific files. - Internal reflection mining: Mines the system's own algorithm reflection logs to surface recurring weaknesses, cross-referencing internal pain with external techniques. - Use Case: Run a weekly upgrade check that scans Anthropic's claude-code changelog and top practitioner blogs, then receive a tiered report (CRITICAL/HIGH/MEDIUM/LOW) of concrete techniques with exact target files and implementation snippets. ## Quick Start Ask the agent to check for upgrades and report new techniques from Anthropic sources relevant to the current system.

Frequently Asked Questions about Upgrade

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check for new Anthropic and Claude Code features automatically?▼

Run the Upgrade workflow, which executes the bundled Bun script to diff 30+ Anthropic sources against stored state. It returns only new items since the last check, each with a priority rating and a recommendation mapped to your system's files.

How does the skill avoid recommending things already implemented?▼

Every recommendation must carry a Prior Status tag backed by file:line evidence gathered during the run. Already-implemented items are moved to a Skipped Content section with proof, and previously rejected ideas only resurface if a named change in context justifies it.

Can I monitor YouTube channels and custom sources for upgrade ideas?▼

Yes. YouTube channels are configured via a youtube-channels.json customization file and checked with yt-dlp, while custom sources are defined in a user-sources.json file. Seen-state is tracked per source so only new videos and pages are processed.

What happens when a source times out during an upgrade scan?▼

The scan enforces a hard deadline of about four minutes and fails open. A source that times out is listed as timed out in the Sources Processed section, degrading coverage for that run without blocking or delaying the final report.

Does the skill only track external sources for improvements?▼

No. It also mines internal algorithm reflection logs for recurring failure patterns and cross-references them with external findings. Recommendations where internal pain and external techniques agree are flagged as the strongest candidates.