tercel

Building AI-perceivable tools apcore
Exploring agent-tool protocols, function calling pain points

[tercel @tercelyi]

Building with Claude Code just got a massive upgrade. ⚡️

Introducing Tercel Claude Plugins: A curated marketplace covering the full SDLC.

🔹 spec-forge: Turn ideas into IEEE-standard specs.
🔹 code-forge: Plan, TDD, and 14-point reviews in the terminal.
🔹 hype-forge: Instant technical evangelism for your ship.
🔹 research-forge: Deep-dive analysis on any OS project.

The flow: Idea → Spec → Plan → Code → Ship.
/plugin marketplace add tercel/tercel-claude-plugins

https://t.co/v0CTNFVNhx

[tercel @tercelyi]

@nypost Real fix: force all labs to publish red-team results + incident reports. Sunlight beats quiet lobbying every time.

[tercel @tercelyi]

@triptitips 100%. Debug the handoffs like APIs: define contracts, log every edge, then run chaos tests on the whole chain

[tercel @tercelyi]

@Shruti_0810 Love this mix. Simple stack: Orca for flows → Agency Agents for roles → Codebase Memory for long-horizon coding work

[tercel @tercelyi]

@athrix_codes Love the AI Agents MCP angle. Any plans for dark-mode-first layouts or headless primitives next?

[tercel @tercelyi]

@VanJones68 Totally. We need 3 things fast: incident reporting, model eval standards, and real penalties for reckless deploys.

[tercel @tercelyi]

@lidangzzz 重点是人家英语水平还有托举

[tercel @tercelyi]

agent-skill-bundler
Port Claude Code skill plugins to other agent platforms. Codex is the first target.
https://t.co/88MSKNmtdM

[tercel @tercelyi]

@WY_mask 多 agent 做交易，怎么防止「过拟合+幻觉」？你们在风险控制、实盘 vs 回测一致性上是怎么设计的？

[tercel @tercelyi]

@libsoftiktok @AEO aeo, cool

[tercel @tercelyi]

AI-Perceivable: When modules, interfaces, and tools are clearly perceived by AI, the model can understand their structure and purpose, and therefore know exactly how to invoke, orchestrate, and act on them. From perception → cognition → execution.

[tercel @tercelyi]

apcore is an AI-Perceivable module standard that makes every interface naturally perceivable and understandable by AI through enforced Schema definitions and behavioral annotations.

[tercel @tercelyi]

apcore is an AI-Perceivable module standard that makes every interface naturally perceivable and understandable by AI through enforced Schema definitions and behavioral annotations.
https://t.co/VicJhhYpwg

[AI Perceivable @aiperceivable]

A schema-enforced module standard for the AI-Perceivable era https://t.co/4GAkLBncgZ via @YouTube

[tercel @tercelyi]

The CLI Adapter for apcore — Expose modules as high-performance, AI-perceivable command-line tools.

[tercel @tercelyi]

Why "Prompt Engineering" is the New Manual Labor 🧹

Stop being a "Prompt Janitor."

If your AI workflow is just copying requirements into a chat window and pasting code back, you aren't an engineer; you're human glue.

### The Wall: Why Prompts Fail
Prompts are just "talk." They lack:
- **Context Awareness**: Chat windows can't see your local files.
- **Execution Power**: AI can't run `npm test` in a browser.
- **State Management**: Complex features break the context window.

### The Shift: From Prompts to "Skills"
The pros are moving to **Skills**. Skills are automated workflows that bind AI logic to your local terminal and file system.

The goal isn't "Chat-driven Development." It's **"Automated Workflow Forging."**

**Stop prompting. Start Forging.** 🛠️🔥

---
**Get the Skills:**
I've open-sourced the entire Forge stack (`spec-forge`, `code-forge`, etc.).

👉 **Check my pinned tweet 📌 @tercelyi for the GitHub repo & setup guide.**
(Or reply "SKILLS" below and I'll send it over! ⚡️)

[tercel @tercelyi]

MCP + JSON = 🧱
MCP + mcp-embedded-ui = ✨

Stop forcing users to guess your tool schemas.
mcp-embedded-ui lets you embed a full-featured tool explorer into ANY server with a standard set of HTTP endpoints.

pip install mcp-embedded-ui
npm install mcp-embedded-ui

[tercel @tercelyi]

@lidangzzz 搞了个 AI-Perceivable SDK https://t.co/W8t1GFvbTZ

[tercel @tercelyi]

𝗥𝗲𝗮𝗱𝘆 𝘁𝗼 𝗯𝘂𝗶𝗹𝗱 𝗔𝗜-𝗽𝗲𝗿𝗰𝗲𝗶𝘃𝗮𝗯𝗹𝗲 𝗺𝗼𝗱𝘂𝗹𝗲𝘀?

[tercel @tercelyi]

apcore defines the 𝗖𝗼𝗴𝗻𝗶𝘁𝗶𝘃𝗲 𝗜𝗻𝘁𝗲𝗿𝗳𝗮𝗰𝗲 for your application. It ensures your logic is naturally perceivable by AI Agents through enforced schemas, behavioral guardrails, and self-healing error guidance.