Public story · 2026-02-17 · source-backed

benchflow-ai/skillsbench — Agent Skills Benchmark (396 stars)

Issue

2026-02-17

Confidence

source-backed

Sources

Redaction

redacted

Story

First standardized benchmark for AI agent skills. 86 tasks, 11 domains, 7,308 test trajectories. Critical finding: curated skills +16.2%, self-generated skills +0%. Run it against your own skills. GitHub | Paper

Skills You Can Learn Today
1. Build a Private Claude Code Plugin Marketplace (intermediate, vibe-coding) — Bundle skills, agents, hooks, MCP servers into installable team plugins via GitHub repos. Docs 2. Google ADK TypeScript Multi-Agent Orchestration (intermediate, agent-patterns) — Code-first agent f...
doop puts a design canvas behind an MCP server
255 stars in the two days after its August 22 creation, AGPL-3.0, positioned against Paper.design. Designs live on a shareable canvas of frames rendering real HTML in sandboxed iframes; people edit in the browser and agents edit through the built-in MCP server, with live curso...
AgentShield: first open benchmark of agent security vendors.
537 test cases across 8 categories testing 6 commercial products. Composite scores range from ~39 to ~98. Critical finding: providers catching >95% of prompt injections miss most unauthorized tool calls — tool abuse detection is universally weak. Provenance verification is nea...
CommerceAgentBench scores agents on what they changed, not what they said.
Alibaba International's Accio team open-sourced 107 tasks (53 CLI, 28 browser, 16 file, 10 API/MCP) running against fourteen offline replicas of real business software in a fresh container per task, with verifiers inspecting mock-service state rather than the transcript. Claud...
garagon/aguara — Static Security Scanner for Agent Skills (16 stars, 3 days old)
GitHub | Go, MIT Extremely new but architecturally significant. Performs static analysis on skill files (markdown, YAML, JSON) to detect threats before deployment — offline, deterministic, no LLM required. 138 detection rules across 15 categories. Companion service scanned 31,...
SkillsBench: Curated Skills Beat AI-Generated Skills 16.2% to Zero
SkillsBench is the first standardized benchmark for evaluating agent skills — 86 tasks across 11 domains, 7,308 test trajectories. The critical finding: curated skills boost agent pass rates by 16.2%, but self-generated skills provide zero benefit on average. Smaller models wi...

Source trail

Entities

Agent Skills Benchmark Critical GitHub Paper

Provenance

Canonical issue: Ramsay Research Agent — 2026-02-17
AI generated: no
Story unit: 2026-02-17-benchflow-ai-skillsbench-agent-skills-benchmark-396-stars
Labels: source-backed, canonical briefing excerpt

Fetching from the wire…

benchflow-ai/skillsbench — Agent Skills Benchmark (396 stars)

Story

Related stories

Skills You Can Learn Today

doop puts a design canvas behind an MCP server

AgentShield: first open benchmark of agent security vendors.

CommerceAgentBench scores agents on what they changed, not what they said.

garagon/aguara — Static Security Scanner for Agent Skills (16 stars, 3 days old)

SkillsBench: Curated Skills Beat AI-Generated Skills 16.2% to Zero

Source trail

Entities

Provenance

benchflow-ai/skillsbench — Agent Skills Benchmark (396 stars)

Story

Related stories

Skills You Can Learn Today

doop puts a design canvas behind an MCP server

AgentShield: first open benchmark of agent security vendors.

CommerceAgentBench scores agents on what they changed, not what they said.

garagon/aguara — Static Security Scanner for Agent Skills (16 stars, 3 days old)

SkillsBench: Curated Skills Beat AI-Generated Skills 16.2% to Zero

Source trail

Entities

Provenance

Story

Related stories

Skills You Can Learn Today

doop puts a design canvas behind an MCP server

AgentShield: first open benchmark of agent security vendors.

CommerceAgentBench scores agents on what they changed, not what they said.

garagon/aguara — Static Security Scanner for Agent Skills (16 stars, 3 days old)

SkillsBench: Curated Skills Beat AI-Generated Skills 16.2% to Zero

Tomorrow's issue, in your inbox.

Source trail

Entities

Provenance

Story

Related stories

Skills You Can Learn Today

doop puts a design canvas behind an MCP server

AgentShield: first open benchmark of agent security vendors.

CommerceAgentBench scores agents on what they changed, not what they said.

garagon/aguara — Static Security Scanner for Agent Skills (16 stars, 3 days old)

SkillsBench: Curated Skills Beat AI-Generated Skills 16.2% to Zero

Tomorrow's issue, in your inbox.

Source trail

Entities

Provenance