Why We Need a “Human Context Protocol” (HCP)

AI Agents are powerful, but in enterprise use, it they often fall short of expectations—largely because open-ended prompts are hard for humans to master at scale.

Model Context Protocol (MCP) helps the model access tools and data—but Human Context Protocol (HCP), a lightweight standard I invented, helps humans access the model more effectively.

⚙️ HCP is simple: Send ? as your only prompt, and the agent replies with quick, actionable shortcuts. Think of it like command-line help for executives.

Example:
Sending ? returns:

plan-day: Build today’s agenda from calendar + tasks
plan-week: Plan your week based on current priorities
escalate : Escalate a support ticket
sales-yesterday: Summarize yesterday’s sales
query-table-range : Run a date-based table query
forecast-cash-180d: Generate a 6-month cash forecast
churn-watch: Highlight accounts at risk
summarize-board-pack : Reduce 60 slides to 5 bullet points

Each shortcut maps to a vetted prompt. It’s MCP in reverse—we reduce friction for the human, not the model.

Why You Should Care:

✅ Immediate ROI
Your prompt libraries already exist—HCP just surfaces them faster.

⚡ Faster Decisions
Replace 10-minute prompt writing with shortcut triggers.

🧠 Lower Cognitive Load
People don’t need to master prompting—just use ?.

🛡️ Governance Built-In
Shortcuts are pre-approved. No more risky shadow prompts.

🔄 Universal UX
Works in chat, voice, email—even AR glasses.

🚀 How to Try HCP:
Catalog Prompts – Identify common, high-value requests
Create Shortcuts – Use hashtag
hashtag#verb-object format (e.g. hashtag
hashtag#plan-week)
Add a Help Trigger – Let ? surface the list
Control Access – Tie shortcuts to user roles
Track and Improve – Measure usage, iterate quarterly
Let’s stop expecting people to engineer prompts—and give them HCP instead.

Curious how HCP could work for your org?
Let’s connect

-David McKee