agent_setting

Use when source code must be implemented, debugged, refactored, or code-audited through a plan, execution, test, and report loop. Not for requirements-only changes, new empirical experiments, or document-only editing.

Use when persistent analysis of code, a paper, or a document must be created or refreshed because it is absent, stale, or explicitly requested. Not for read-only project orientation, context recovery, or status reporting.

Use when durable cross-project user preferences must be inferred from coding, writing, or analysis evidence and stored as a profile. Not for one-project context recovery, casual preference acknowledgment, or ordinary task execution.

Use when a visual product surface needs references, design tokens, components or mockups, review, and development handoff. Not for implementing an already-approved design in code or for document prose work.

Use when a new paper, presentation, report, proposal, or other user-facing document must be produced from evidence. Not for correcting only an existing document or for source-code implementation.

Use only when autopilot-draft dispatches document strategy and evidence-plan creation. Not for top-level user requests or primary capability routing.

Use only when autopilot-code dispatches the implementation stage for an approved plan. Not for top-level user requests or primary capability routing.

Use only when autopilot-code dispatches the planning and plan-check stage. Not for top-level user requests or primary capability routing.

Use only when autopilot-code dispatches plan revision after user notes, plan-check feedback, or verification failure. Not for top-level user requests or primary capability routing.

Use only when autopilot-code dispatches the final code-cycle reporting stage. Not for top-level user requests or primary capability routing.

Use only when autopilot-code dispatches implementation verification and evidence recording. Not for top-level user requests or primary capability routing.

Use only when autopilot-design dispatches component, mockup, or preview construction. Not for top-level user requests or primary capability routing.

Use only when autopilot-design dispatches visual-reference collection and brief creation. Not for top-level user requests or primary capability routing.

Use only when autopilot-design dispatches design quality, token-contract, and breakage review. Not for top-level user requests or primary capability routing.

Use only when autopilot-design dispatches design-token definition or revision. Not for top-level user requests or primary capability routing.

Use only when autopilot-draft or autopilot-refine dispatches an internal strategy or draft refinement stage. Not for top-level user requests or primary capability routing.

Use when completed work or artifacts need a read-oriented inspection for drift, inconsistency, omissions, or unsupported claims. Not for implementing fixes, producing the primary artifact, or replacing execution-stage verification.

Use when training setup, checkpoint evaluation, ablation, metric computation, or other new empirical output is required. Not for ordinary production-code changes that create no new experiment or evaluation result.

Use when existing artifacts must be collected, digested, triaged, or routed into a durable note workflow. Not for storing one scoped memory item or for performing the primary research, code, or document work.

Use when an existing document or research artifact needs factual, structural, stylistic, or review-driven correction with history preserved. Not for drafting a new artifact or performing new empirical work required before the document can change.

Use when a task needs a durable survey of new academic, technology, or market evidence before downstream specification or production. Not for repository-only project analysis, a simple factual lookup, or work already grounded by sufficient current evidence.

Use when an application needs deployment or release configuration, readiness verification, and a ship checklist. Not for ordinary feature implementation or an unauthorized production deployment.

Use when product requirements, architecture, evaluation policy, or another blueprint must be created or materially updated before implementation. Not for implementing an already-approved specification or for editing unrelated documents.

Use only when autopilot-code dispatches implementation verification and evidence recording. Not for top-level user requests or primary capability routing.

Use when an approved cheatsheet or patching guide must be applied to the real source artifact and the applied result verified. Not for drafting or refining the cheatsheet, or for unrelated source implementation.

Use when persistent analysis of code, a paper, or a document must be created or refreshed because it is absent, stale, or explicitly requested. Not for read-only project orientation, context recovery, or status reporting.

Use when durable cross-project user preferences must be inferred from coding, writing, or analysis evidence and stored as a profile. Not for one-project context recovery, casual preference acknowledgment, or ordinary task execution.

Use when completed work or artifacts need a read-oriented inspection for drift, inconsistency, omissions, or unsupported claims. Not for implementing fixes, producing the primary artifact, or replacing execution-stage verification.

Use when an approved cheatsheet or patching guide must be applied to the real source artifact and the applied result verified. Not for drafting or refining the cheatsheet, or for unrelated source implementation.

Use when source code must be implemented, debugged, refactored, or code-audited through a plan, execution, test, and report loop. Not for requirements-only changes, new empirical experiments, or document-only editing.

Use when a visual product surface needs references, design tokens, components or mockups, review, and development handoff. Not for implementing an already-approved design in code or for document prose work.

Use when a new paper, presentation, report, proposal, or other user-facing document must be produced from evidence. Not for correcting only an existing document or for source-code implementation.

Use when training setup, checkpoint evaluation, ablation, metric computation, or other new empirical output is required. Not for ordinary production-code changes that create no new experiment or evaluation result.

Use when existing artifacts must be collected, digested, triaged, or routed into a durable note workflow. Not for storing one scoped memory item or for performing the primary research, code, or document work.

Use when an existing document or research artifact needs factual, structural, stylistic, or review-driven correction with history preserved. Not for drafting a new artifact or performing new empirical work required before the document can change.

Use when a task needs a durable survey of new academic, technology, or market evidence before downstream specification or production. Not for repository-only project analysis, a simple factual lookup, or work already grounded by sufficient current evidence.

Use when an application needs deployment or release configuration, readiness verification, and a ship checklist. Not for ordinary feature implementation or an unauthorized production deployment.

Use when product requirements, architecture, evaluation policy, or another blueprint must be created or materially updated before implementation. Not for implementing an already-approved specification or for editing unrelated documents.

Use when persistent analysis of code, a paper, or a document must be created or refreshed because it is absent, stale, or explicitly requested. Not for read-only project orientation, context recovery, or status reporting.

Use when durable cross-project user preferences must be inferred from coding, writing, or analysis evidence and stored as a profile. Not for one-project context recovery, casual preference acknowledgment, or ordinary task execution.