Problem / Friction
Post-v13, roadmap planning lapsed (operator, 2026-06-15: "something we neglected post the v13 release"). Empirically: no release milestone exists, and ~30 open epics span harness + Body + Brain with no release boundary telling the team which are now vs abstract — the "abstract goals to chase" friction. A release shipped, but nothing fired to (a) mark the win or (b) scope the next roadmap. There is no repeatable discipline for the post-release planning beat.
This is the creation-side dual of the friction the v13.1-scoping exercise surfaced: a release should trigger the next-roadmap, not leave it to chance.
Intent
A repeatable update-roadmap agent skill that fires post-release:
- Celebrate first — acknowledge the shipped release (the win; reward-continuity for the swarm).
- Plan the next roadmap — define the next release milestone's scope: the cornerstones + the rationale (why / intent), NOT an exhaustive item-list. A roadmap names the load-bearing direction, not every ticket. Scope which epics/discussions are in (clarity) vs deferred (no abstract-chasing). Concept-wise: a GitHub release milestone that scopes a set of epics/discussions + a short "why this release" rationale.
Skill shape (for the author — built via /create-skill)
- Trigger: a release ships (vX.Y merged to
main).
- Steps: celebrate → assess what landed → define the next release milestone → scope in / defer epics + discussions → assign them to the milestone → record the cornerstones + rationale.
- Output: a release milestone with a scoped epic/discussion set + a short rationale; the exhaustive set is the milestone's linked items (queried, not listed in prose — mirrors
epic-create's subs-linked-not-listed rule).
Avoided traps / rejected shapes
- Roadmap = exhaustive item-list — rejected; cornerstones + rationale only. The full set is the milestone's linked epics/discussions, queried on demand.
- Skipping the celebration — rejected; the win + reward-continuity matter for the swarm (an institution goal).
- A roadmap with no explicitly-deferred set — rejected; the clarity comes from a visible out, not just in (else deferred items stay silently-abstract).
Relationship
- Adjacent to #10321 ("Author release-workflow agent skill", deferred-until-v12.2 — that condition is now stale: v13 shipped). #10321 is release-execution (how to cut a release); this is post-release planning. They are siblings in a release-lifecycle skill cluster — consider co-scoping, or reviving #10321 under the same cluster.
- The skill is loaded substrate (
.agents/skills/**) → it MUST be built via /create-skill (Map-vs-World-Atlas: a trigger line in the always-loaded Map, the rule body in references/) + /turn-memory-pre-flight (load-runtime-effect audit).
Authored by @neo-opus-vega (Vega), harness steward.
Problem / Friction
Post-v13, roadmap planning lapsed (operator, 2026-06-15: "something we neglected post the v13 release"). Empirically: no release milestone exists, and ~30 open epics span harness + Body + Brain with no release boundary telling the team which are now vs abstract — the "abstract goals to chase" friction. A release shipped, but nothing fired to (a) mark the win or (b) scope the next roadmap. There is no repeatable discipline for the post-release planning beat.
This is the creation-side dual of the friction the v13.1-scoping exercise surfaced: a release should trigger the next-roadmap, not leave it to chance.
Intent
A repeatable
update-roadmapagent skill that fires post-release:Skill shape (for the author — built via
/create-skill)main).epic-create's subs-linked-not-listed rule).Avoided traps / rejected shapes
Relationship
.agents/skills/**) → it MUST be built via/create-skill(Map-vs-World-Atlas: a trigger line in the always-loaded Map, the rule body inreferences/) +/turn-memory-pre-flight(load-runtime-effect audit).Authored by @neo-opus-vega (Vega), harness steward.