Why CEREBRO kept it

Paper assessing Claude Code skill effectiveness; harness evaluation

Backlinks

Appeared in 1 briefing

• 2026-11-04

Related

Shares tags: ai/agents · cerebro/signal

• Each agent builds up a wiki over time so I can drop the conversation history and switch harnesses as required Plus, each one knows about my skill pack and can run /implement-spec or /retro as neededx · 2026-11-04
• 'Harness self-evolution requires more than generating and implementing changes: it needs ways to establish whether a change helps on later tasks and to revise or remove changes that do not.' In otherx · 2026-11-04
• Thinking about trialling a combo of two approaches: - A local, long-running orchestrator thread covering a single domain (maybe with my /chief-of-staff skill) - Attached to a software factory, with sx · 2026-11-04
• An update on building games with Claude Code: Last week some of you ran into Claude Code pushing back on game projects it should have just built. We'd added a short rule to the Opus 5.5 prompt to stex · 2026-11-04
• Your agent returned the right answer. How many attempts did it take to get there? A retry adds a model call. A slow dependency adds another tool attempt. More context adds tokens. A larger model addsx · 2026-11-04
• kwantrss · 2026-11-04

Also from x.com

• Stageshow might be the best free skill for launch videos I've seen Without it, the UI looks like a pasted screenshot. With it, you get an actual STORY. 22,627 bookmarks in 37 hours. Benji Taylor gax · 2026-11-04
• Each agent builds up a wiki over time so I can drop the conversation history and switch harnesses as required Plus, each one knows about my skill pack and can run /implement-spec or /retro as neededx · 2026-11-04
• 'Harness self-evolution requires more than generating and implementing changes: it needs ways to establish whether a change helps on later tasks and to revise or remove changes that do not.' In otherx · 2026-11-04
• Thinking about trialling a combo of two approaches: - A local, long-running orchestrator thread covering a single domain (maybe with my /chief-of-staff skill) - Attached to a software factory, with sx · 2026-11-04
• An update on building games with Claude Code: Last week some of you ran into Claude Code pushing back on game projects it should have just built. We'd added a short rule to the Opus 5.5 prompt to stex · 2026-11-04
• Users can now auto-connect plugins when they log in to Claude, with MCP servers that accept Claude as a trusted issuer. If you’re building plugins for Claude, add support so users get started fasterx · 2026-11-04