How the i‑have‑adhd Plugin Handles "Explain" and "Walk Me Through" Requests

The i‑have‑adhd plugin detects when a user asks to "explain" or "walk me through" a topic and temporarily overrides its default brevity rules to generate a full, multi‑paragraph explanation with structural headers while still omitting preambles and closers.

The i‑have‑adhd repository provides a specialized AI assistant skill designed for terse, action‑first communication. When users need deeper context, the primary keyword—handling requests to explain or walk me through—triggers a specific rule‑override documented in the skill definition files.

Rule Override Mechanism

The plugin normally enforces strict output‑shaping rules that minimize verbosity. However, the architecture includes an explicit escape hatch for educational requests.

The Default Communication Rules

By default, the skill applies constraints that produce concise, scannable output:

• No preambles or closers – responses start immediately with content and end without sign‑offs.
• Numbered steps – actions are presented as sequential tasks.
• Minimal length – explanations are kept brief to reduce cognitive load.

These rules are defined in skills/i-have-adhd/SKILL.md and repeated in skills/i-have‑adhd/agents/gemini.toml.

Detecting Explain/Walk‑Through Intent

When the user’s intent matches "explain" or "walk me through", the plugin activates a documented exception found under the section "When to break the rules". According to the source code in SKILL.md, the override instructs the engine to:

"Explain fully. Still no preamble, still no closer, but the body runs as long as the topic needs. Add headers so the reader can skim back."

The same conditional logic appears in gemini.toml, confirming the behavior is consistent across agent configurations: "Break these rules only when: the reader asks you to 'explain' or 'walk through' (go as long as the topic needs, still no preamble/closer)."

Response Structure Changes

Activating the explain mode modifies how the output formatter assembles the final response.

Content Length and Headers

The most significant change is the removal of length constraints. While default responses truncate elaboration, explain mode allows the body to continue until the subject is completely covered. To maintain scannability despite the increased length, the plugin inserts markdown headers (## Section Name) throughout the text. This creates anchor points that let readers skim backward to specific sections of interest.

Persistent Style Constraints

Even in explain mode, the plugin preserves two critical style rules:

1. No greeting – the response begins directly with the first header or paragraph.
2. No closing pleasantry – the text ends immediately after the final point without recap or sign‑off.

Configuration Sources

The behavior is hardcoded in two primary locations within the repository:

• skills/i-have-adhd/SKILL.md – Contains the canonical skill definition, including the "When to break the rules" section that defines the explain/walk‑through override.
• skills/i-have-adhd/agents/gemini.toml – Agent‑specific configuration that replicates the same override wording for the Gemini harness.

Both files explicitly link the trigger phrases "explain" and "walk me through" to the relaxed output constraints.

Practical Usage Examples

Below are minimal examples showing how to invoke the explain mode.

Trigger an explanation using the "explain" keyword:

/i-have-adhd

Explain how OAuth 2.0 works.

Resulting behavior: The response omits any greeting, provides comprehensive multi‑paragraph coverage of OAuth 2.0 flows, uses headers like # Overview and # Authorization Code Grant, and ends abruptly after the final technical point.

Trigger the same override with alternative phrasing:

Walk me through setting up a new GitHub Actions workflow.

Resulting behavior: The plugin generates an extensive step‑by‑step guide with structural headers for each phase (e.g., ## Creating the YAML File, ## Defining Triggers), maintaining the no‑preamble/no‑closer style while allowing the content to run as long as necessary.

Summary

• The i‑have‑adhd plugin enforces terse, action‑first rules by default but applies a rule‑override for explain requests.
• When the user says "explain" or "walk me through," the body may run as long as the topic requires.
• Structural headers are automatically added to long explanations to aid navigation.
• The no‑preamble and no‑closer constraints remain active even in explain mode.
• Configuration resides in SKILL.md and gemini.toml under "When to break the rules".

Frequently Asked Questions

Does the plugin completely disable all formatting rules for explain requests?

No. While the length constraints are relaxed, the plugin still enforces the no‑preamble and no‑closer rules. The response will never start with a greeting like "Sure, I can help" or end with "Let me know if you need anything else." It begins directly with content and ends immediately after the final technical detail.

Where is the explain/walk‑through rule override documented?

The override is documented in skills/i-have-adhd/SKILL.md under the section "When to break the rules" → "1. User asks to 'explain' or 'walk me through.'" The same logic is mirrored in skills/i-have-adhd/agents/gemini.toml, ensuring consistent behavior across different agent harnesses.

Can I use "walk me through" and "explain" interchangeably?

Yes. According to the source configuration in gemini.toml and SKILL.md, both phrases trigger the same rule suspension. Whether you ask the assistant to "explain recursion" or "walk me through recursion," the plugin identifies the educational intent and switches to the long‑form, header‑rich output mode.

Does the plugin add a table of contents when explaining complex topics?

No. While the plugin adds markdown headers (## Section) to break up long explanations, it does not generate an explicit table of contents or index. The headers alone provide visual landmarks for skimming, consistent with the skill's preference for minimal overhead.

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