Descriptions:
Engineers… the future is built by those who PLAN it, not those who vibe code it. 🔥 The recently banned Fable 5 and Mythos class models just unlocked a whole new level of planning, and great planning IS great engineering. Most engineers hand their /plan off to the model and pray. That’s the mass deprecation of raw engineering talent in real time.
🎥 VIDEO REFERENCES
– GET THE /PLANF3 SKILL: https://github.com/disler/planf3
– Tactical Agentic Coding: https://agenticengineer.com/tactical-agentic-coding?y=DzbqeO_diOQ
– PI to PI Agent Communication: https://youtu.be/PIdETjcXNIk
– Fable 5: https://www.anthropic.com/news/claude-fable-5-mythos-5
– Fable 5 Banned: https://www.anthropic.com/news/fable-mythos-access
– Iroh: https://docs.iroh.computer/what-is-iroh
– GPT Image 2: https://openai.com/index/introducing-chatgpt-images-2-0/
This is a slow, in-depth agentic engineering devlog. If you’re waiting for a get-rich-quick vibe coding hack or a flashy multi-agent demo, click away now. This is for agentic engineers who know that knowing what your agents are doing is the key to success at scale.
🚀 In this video, IndyDevDan rebuilds the /plan meta skill from the ground up for Fable 5 and the new Mythos class models. We start where all great work starts: sit down, type it out, and think. Then we bring the agent in to template our engineering into a reusable plan format we’ll use hundreds of times. This is planning, agentic engineering, and claude code workflow built for the next class of mythos model intelligence.
🛠️ Why rebuild the plan skill at all? Because your planning skill is the single most important tool you and your agents have. It decides how good your results are BEFORE a single file gets modified. We push hard against the two real constraints of agentic coding — planning and reviewing — by investing more up front. More planning means less reviewing, and that gap only widens as every new mythos class model ships.
💡 The trade-off trifecta is performance, speed, cost. This new plan template spends speed and tokens to win on performance. We’re not optimizing for cheap. We scale our compute to scale our impact.
🔥 What we build into the new /plan F3 (PLANS For Fable 5) skill:
• HTML-first plan output — give your agent more useful tokens, easier for you, your team, and your AI agents to consume
• Embedded GPT Image 2 generation — one focused image per phase so a plan is worth a thousand words
• A toggleable questions-and-answers section — step into the loop only when the work is intense enough
• Rich, updatable header metadata — created, modified, commits, agent name, session id, forward and back references
• Per-phase and per-task breakdowns with an embedded checklist your agents update live
• Validation and testing sections with closed loops that block completion until every box is checked
• Dedicated workflows — create plan, update plan, update references, build plan, and image generation, all in one unified skill
⚡ The big idea behind this devlog: we treat the plan as a living artifact in the codebase, written and read by the agent trifecta — you, your engineering team, and your AI agents. We’re not building only for the model. By templating our engineering, we teach the agent how WE engineer, and that’s what pulls your results out of the normal distribution.
🎯 Then we put it to work — using my pi coding agent and agent-to-agent communication, we point the brand new plan skill at re-implementing my network layer with Iroh and watch it ship a complete, image-rich HTML spec. Running on Opus 4.8 today, this gets you 80-90% of the way there. Run it on a true mythos class model and the results get cracked.
Domain knowledge is what differentiates you. Encode it into your skills, template your engineering, and stop outsourcing your thinking.
Stay focused and keep building.
IndyDevDan
#agenticengineering #claudefable #agenticcoding
