AI Framework Watch

Generates weekly competitive-intelligence digests tracking releases, stars, and breaking changes across nine AI agent frameworks via the GitHub API.

Updated May 25, 2026
One-click install
npx skills add https://github.com/Cbowcrptex/CBOWCRYPTEX --skill ai-framework-watch-cbowcrptex
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: AI Framework Watch
Source: https://github.com/Cbowcrptex/CBOWCRYPTEX/tree/main/skills/ai-framework-watch
Command: npx skills add https://github.com/Cbowcrptex/CBOWCRYPTEX --skill ai-framework-watch-cbowcrptex

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Operators building on AI agent frameworks must manually read changelogs across many repositories every week to catch breaking changes, momentum shifts, and release activity, which is time-consuming and easy to fall behind on. ## Core Features & Use Cases - Watchlist sweep: Tracks a hardcoded set of 9 frameworks (aeon, LangGraph, CrewAI, AutoGen, LlamaIndex, Mastra, smolagents, dspy, Pydantic AI) using gh api for stars, forks, issues, and releases. - Breaking-change detection: Flags releases with explicit breaking markers, major-version bumps, or migration language, and computes 7-day and 30-day star deltas from persisted state. - Deep-dive mode: Passing a framework slug produces a single-framework report with 30 days of releases, top reacted issues, and top contributors. - Use Case: Every Monday, run the sweep to get a ranked table, a one-line verdict (e.g. BREAKING WEEK or MOMENTUM SHIFT), a Markdown article, and a gated notification so build decisions like adopting a LangGraph migration are grounded in current data. ## Quick Start Run the AI Framework Watch skill to generate this week's digest of releases, star momentum, and breaking changes across the tracked AI agent frameworks.

Frequently Asked Questions about AI Framework Watch

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

FAQPage Schema
How do I track breaking changes across AI agent frameworks weekly?▼

Run the sweep mode, which pulls the last 5 releases per framework via gh api, filters to the last 7 days, and flags releases containing markers like BREAKING:, major-version bumps, or migration language. Flagged releases surface in the verdict line and digest article.

How do I get a deep dive on a single framework like LangGraph?▼

Pass the framework slug as the optional variable, for example langgraph. The skill switches to deep-dive mode and writes a report with 30 days of releases, the top 3 open issues by reactions, and the top 3 contributors over the last 90 days.

Which AI agent frameworks are tracked by the watchlist?▼

The hardcoded watchlist covers nine frameworks: aeon, LangGraph, CrewAI, AutoGen, LlamaIndex, Mastra, smolagents, dspy, and Pydantic AI. Adding or removing a framework requires an explicit edit to the SKILL.md file to preserve week-over-week comparability.

Does this skill require a GitHub API key or token setup?▼

No manual token configuration is needed. The skill authenticates through the GitHub CLI using the GITHUB_TOKEN environment variable automatically, and all data access is read-only through gh api with no curl or custom headers.

What happens when the GitHub API rate-limits or a repo is unreachable?▼

Rate limits trigger one retry after 60 seconds, and persistent failures mark the framework unreachable for that run. If fewer than 3 frameworks are reachable the run exits with an error status and skips state mutation so future deltas are not corrupted.

When does the skill skip sending a notification?▼

Notifications are skipped when the run is quiet, meaning the verdict is STEADY with no breaking flags, no momentum picks, and it is not the first run. The result is still written to the memory log and article file.