Against Skill sprawl: why 50 saved prompts beat 50 installed Skills
TruthTellerTom · reddit · 2026-10-31
A Reddit user argues for deliberately installing few Skills and leaning on saved prompts instead:
- Only 5-6 genuinely needed Skills installed; everything else uses saved prompts, custom instructions, and reusable templates.
- Explicit beats implicit: invoke the exact prompt from a menu instead of hoping the model discovers the right Skill.
- Skills add a decision point: the model must evaluate what's available, risking missed, ignored, or wrongly-triggered Skills.
- More Skills = more noise: descriptions and discovery mechanisms still add context overhead, with overlapping instructions as collections grow.
- Saved prompts are portable: the same prompt works across OpenCode, Codex, IDEs, and web chats; editing one place updates everywhere.
- Control over triggering: the human decides when a workflow runs.
The author concedes Skills still shine for packaging specialized knowledge, scripts, and complex repeatable workflows.
More from coding & agent
- Vercel Labs ships agent-desktop: token-efficient computer use for the whole OS — cramforce · 2026-10-31
- 83 AI agents spend 45 hours and 3.9B tokens creating "what SI thinks about alone" — tetsuoai · 2026-10-31
- GitHub Copilot App v1.1.34 ships 117 features: faster session resume, agent merge, CSV tables — 0xkarasy · 2026-10-31
- Playwright MCP Server lets AI assistants drive remote browsers — modelcontextprotocol · 2026-10-31
- zero-core-tools: MCP connector bundling scraping, code review, sentiment — modelcontextprotocol · 2026-10-31
- Codex power user shows 71.5B lifetime tokens and a 262-day daily streak — TheMoonMidas · 2026-10-31