Ending a Response with a Concrete Next Action: The i-have-adhd Rule Explained

Every response must conclude with a single, concrete next step that the user can execute in under two minutes.

The i-have-adhd repository provides a strict communication framework designed for neurodivergent workflow management, with ending a response with a concrete next action serving as a mandatory output constraint. This rule appears in multiple canonical documents and defines the final line of every generated response as a time-bounded, executable instruction.

The Canonical Definition in SKILL.md

The primary specification lives in skills/i-have-adhd/SKILL.md, which mandates: "End with one concrete next action." This directive is part of a comprehensive ADHD-friendly style guide that also requires leading with the next action, numbering multi-step tasks, and giving explicit time estimates. The file establishes this as a non-optional formatting requirement for all compliant responses.

According to the source, this rule works alongside other constraints including restating state across turns, suppressing tangents, and making wins visible.

The Two-Minute Execution Constraint

Supporting documentation tightens the scope of what constitutes "concrete." According to INSTALL.md, the rule specifically requires ending with "one next action doable in under two minutes." This time limit prevents decision paralysis by ensuring the recommended step feels immediately achievable.

The README.md reinforces this by emphasizing that every response must finish with a specific, bounded action rather than vague suggestions or open-ended questions.

Implementation Examples

The repository provides clear templates for compliant and non-compliant outputs.

A correct implementation follows this structure:

✅ Summary: Your project has three pending tasks.
1️⃣ Update the README to reflect the new feature.
2️⃣ Add unit tests for the new endpoint.
3️⃣ Run the test suite locally.

⏱ Estimated effort: ≈ 5 minutes total.

👉 Next action: Open `README.md` and add a "New Feature" section (≈ 2 minutes).

An incorrect implementation fails by omitting the final concrete step:

✅ Summary: Your project has three pending tasks.
1️⃣ Update the README to reflect the new feature.
2️⃣ Add unit tests for the new endpoint.
3️⃣ Run the test suite locally.

⏱ Estimated effort: ≈ 5 minutes total.

(Note: No explicit next step provided.)

Reference Files and Enforcement

This rule is enforced across four key files in the repository:

• skills/i-have-adhd/SKILL.md: Contains the canonical definition "End with one concrete next action"
• README.md: Emphasizes the ADHD-friendly response structure including the final concrete action requirement
• INSTALL.md: Specifies the under-two-minute constraint for the closing action
• .opencode/command/i-have-adhd.md: Mirrors these formatting rules for OpenCode command integration

Summary

• Every response must end with exactly one concrete next action, as defined in skills/i-have-adhd/SKILL.md
• The action must be executable in under two minutes according to INSTALL.md
• Vague conclusions or missing final steps violate the protocol
• Correct implementations use a clear formatting indicator followed by a specific, time-bounded instruction

Frequently Asked Questions

What exactly constitutes a "concrete" next action?

A concrete next action is a specific, observable task with clear boundaries. According to the source documentation in skills/i-have-adhd/SKILL.md, it must be a single step that requires no further planning or decision-making to begin. Examples include "Open config.json and add the debug flag" or "Run npm test in your terminal."

Why is there a two-minute time limit on the final action?

The two-minute constraint, specified in INSTALL.md, prevents task overwhelm and reduces activation energy required to start. By ensuring the next action is small enough to complete immediately, the framework helps users maintain momentum and avoid procrastination associated with large, undefined workloads.

Can a response end with multiple next actions or options?

No. The rule explicitly requires "one concrete next action" (singular). While the body of the response may list multiple steps or a numbered sequence, the final line must isolate a single starting point. This prevents decision paralysis and aligns with the ADHD-friendly principle of reducing cognitive load at transition points.

Where is this rule enforced in the codebase?

The rule is documented in skills/i-have-adhd/SKILL.md as the canonical skill definition, referenced in README.md for user guidance, and reiterated in INSTALL.md with the time constraint. Additionally, .opencode/command/i-have-adhd.md applies these same formatting standards when the skill is invoked as an OpenCode command.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}