View on GitHub

campaign-mode

Work with a party of AI advisors with genuinely different perspectives.

ADR-CM-009: Quest Entry Commands

Field Value
Decision ID ADR-CM-009
Type Child ADR
Parent ADR ADR-CM-001
Initiative Campaign Mode
Proposed By Chris Barlow
Date 2026-02-14
Status Accepted

WH(Y) Decision Statement

In the context of extending Campaign Mode with ergonomic entry-point commands,

facing users needing to remember agent-specific slash commands to start or resume campaigns — creating unnecessary cognitive load and breaking the proactive elicitation principle established in ADR-CM-008,

we decided for two dedicated commands (/start-quest and /continue-quest) that delegate to NPC agents and present natural-language options via AskUserQuestion,

and neglected a single overloaded /quest command (ambiguous intent — start vs continue?) and auto-routing without user input (violates the User as Protagonist principle),

to achieve zero-knowledge campaign entry where users can begin or resume quests without memorising agent-specific commands, natural re-entry after breaks with context-aware next-step options, and consistency with the proactive elicitation pattern,

accepting that two additional commands must be maintained alongside the existing agent commands, and direct agent invocation (/gandalf-agent, /dragon-agent, /guardian-agent) remains valid for users who prefer it.


Context

After /campaign-setup completes, users are told to invoke /gandalf-agent to begin a quest. Mid-campaign, users returning after a break must remember which agent to invoke next. Both scenarios require command vocabulary knowledge that the proactive elicitation principle (ADR-CM-008) was designed to eliminate within agent interactions — but the gap remains at campaign entry and re-entry points.

Two new commands address this:

Both commands follow the proactive elicitation pattern: natural language options via AskUserQuestion, no slash command references in user-facing text.

Options Considered

Option 1: Two Dedicated Commands (Selected)

Separate /start-quest and /continue-quest commands with clear, distinct purposes.

Pros: Unambiguous intent, each command does one thing well, consistent with proactive elicitation Cons: Two more commands to maintain

Option 2: Single /quest Command (Rejected)

One overloaded command that auto-detects whether to start or continue.

Why rejected: Ambiguous — when a quest exists, should it start a new one or continue? Forces the command to guess intent, which undermines user agency.

Option 3: Auto-Routing (Rejected)

Automatically detect campaign state and invoke the appropriate agent without user input.

Why rejected: Violates the User as Protagonist principle. The user should choose what to do next, not have the system decide for them.


Dependencies

Relationship ADR ID Title Notes
Extends ADR-CM-008 Proactive Elicitation Applies proactive elicitation to campaign entry/re-entry points
Refines ADR-CM-001 Campaign Mode Architecture Adds ergonomic entry commands to the campaign lifecycle
Relates To ADR-CM-007 Plugin-Based Distribution New commands distributed as part of the plugin

References

Reference ID Title Type Location
REF-001 Campaign Lifecycle Spec Internal Spec SPEC-CM-001-B
REF-002 Campaign Guidelines Spec Internal Spec SPEC-CM-007-B
REF-003 Proactive Elicitation ADR Internal ADR ADR-CM-008

Governance

Review Board Date Outcome Action Review Cadence Next Review
Quarterly

Status History

Status Approver Date
Proposed Chris Barlow 2026-02-14
Accepted Chris Barlow 2026-02-14