mrkeyoor.com_
Mon 10 Aug 08:23 UTC
Open Source10 Aug 2026 05:00 UTC3 min read

Velocity Report: the 10 fastest-growing repos we track (week of 10 Aug 2026)

This week's velocity is dominated by AI agent infrastructure. TencentDB's agent memory tool leads with +6,477 stars, followed by cybersecurity and AI gateway tools.

TencentDB Agent Memory gained 6,477 stars this week, leading our velocity report. The project, a self-hosted memory hub for AI agent teams, has now reached a total of 18,823 stars.

This report is based on MrKeyoor's own week-over-week star tracking across more than 300 open-source repositories we have reviewed.

The board — week of 10 Aug 2026

# Repo Stars gained Total stars
1 TencentDB-Agent-Memory +6,477 18,823
2 reverse-skill +5,989 22,197
3 OmniRoute +5,302 43,250
4 ai-agent-book +4,688 34,712
5 Agent-Reach +4,088 68,625
6 orca +4,041 40,987
7 taste-skill +3,906 74,163
8 ponytail +3,848 98,872
9 firecrawl +3,817 164,236
10 firecrawl +3,771 163,229

Why they are moving

1. TencentDB-Agent-Memory — It solves the critical problem of amnesia in AI agent teams by providing a self-hosted, shared memory hub.

2. reverse-skill — This framework provides structured cybersecurity playbooks, offering a practical guide for security professionals navigating complex threats.

3. OmniRoute — It simplifies development by consolidating hundreds of AI model providers into a single, unified API endpoint.

4. ai-agent-book — This open-source book demystifies AI Agents, offering a deep dive into engineering principles for developers.

5. Agent-Reach — Developers are using this utility to give AI agents live internet access, abstracting away complex connection logic.

6. orca — Orca acts as a specialized IDE for AI coding agents, solving the messy problem of managing agent workflows.

7. taste-skill — It provides structured prompts that help developers improve the design quality and consistency of AI-generated frontends.

8. ponytail — This popular plugin for AI coding assistants helps developers by guiding models to write simpler, more efficient code.

9. firecrawl — Its API is gaining traction by turning messy, unstructured web data into clean, usable formats for AI applications.

10. firecrawl — This API-first service is growing by efficiently scraping and converting complex websites into AI-ready data for developers.

The pattern

This week's data shows a strong developer focus on AI agent infrastructure and tooling. Seven of the top ten projects directly address challenges in building, managing, or augmenting AI agents, from memory and internet access to specialized IDEs and data ingestion.

Methodology: mrkeyoor.com snapshots star counts for every repo it has reviewed and computes week-over-week deltas from baselines at least 5 days old. Full reviews with scorecards: the repo board.

We reviewed this

  1. agents — our honest review
  2. firecrawl — our honest review
  3. taste-skill — our honest review

Sources

  1. MrKeyoor star-tracking data (own dataset)