pulse

Use after completed Pulse work when the user asks to capture post-cycle machine-readable learnings for future planning and execution.

Use when the user asks to distill accumulated dev-notes into reader-facing topic knowledge, rather than to capture a new note or update machine memory.

Use when implementing approved Pulse beads, either as a swarm worker or as a single worker when swarming is unavailable.

Use before implementation planning when feature intent is clear but implementation decisions are still unresolved, and the user needs those decisions locked first.

Use after pulse:exploring when locked decisions are ready and the user needs a mode-gated plan with an approved work shape and current-work prep for validation.

Use when starting any Pulse workflow, resuming a Pulse session, or before planning or execution when tool readiness or onboarding state may block the flow.

Use after execution completes when the user needs final quality verification and release-readiness review before closeout.

Use after pulse:validating approves execution and swarm mode is recommended, when the current work slice should be run by coordinated parallel workers.

Use when bootstrapping or resuming work in a Pulse project after pulse:preflight, or when a request needs Pulse work-shape/current-work routing and mode-aware skill selection.

Use after planning and before execution to verify approved current work is reality-fit, feasibility-ready, and explicitly approved at Gate 3.

Use when starting work in an unfamiliar repository, resuming after lost context, or when the user asks to bootstrap repo understanding before implementation.

Use before feature work only when the design is still unclear — for shaping vague product intent, comparing possible feature shapes, and turning fuzzy requests into a documented, approved spec before exploring or planning begins. Do not use this when the feature intent is already clear and the remaining work is to lock implementation decisions; that belongs to pulse:exploring. Trigger phrases: design, brainstorm, think through, what should we build, help me think, explore options, compare directions, unclear feature shape, vague product request.

Use when Pulse planning or discovery needs graph-based architecture context, execution-flow tracing, symbol relationships, or blast-radius analysis beyond plain file search.

Use when the user asks to improve, structure, or templatize a prompt before execution by Codex or another AI agent.

Use when repository documentation is stale and the user asks to refresh docs to match the current codebase and workflows.

Use when the user asks for root-cause-first debugging of bugs, test failures, flaky behavior, build issues, or integration defects with tracked verification and regression lock-down.

Use when creating a new pulse skill, editing an existing pulse skill, or verifying a skill works under pressure before deploying. Use when you need an agent skill to be bulletproof against rationalization. Do NOT use for project-specific AGENTS.md conventions or one-off solutions.

Repo-wide or subsystem-wide architecture hygiene pass that surfaces deepening opportunities, shallow modules, leaky seams, and ownership drift. Default mode is report-only: no bead creation or execution unless explicitly requested.

Use only when the user explicitly asks to save, note, record, or preserve one learning from the current conversation as a raw developer note.

Use when the user asks for a manual consolidation pass over runtime artifacts into machine-readable Pulse memory outside the post-cycle compounding flow.