flower
/
All briefs
planned draft note flower

Verify direct-dispatch v1 (#36) works end-to-end

Dispatch

canonical · plan

Spec

markdown

hand-off · dispatch

Dispatch

Auto-dispatch

when it reaches planned

Design-loop

design pass before build

kind

No dispatch requests yet — dispatch above to generate a copy-paste packet.

provenance · append-only

Trace

live
or paste a screenshot uploading…
  1. link added 1d ago
    agent · flower-refine
  2. unblocked 1d ago

    Unblocked — #36 reached complete.

    system · flower-36
  3. participant joined 1d ago
    system · flower-36
  4. link added 2d ago
    agent · flower-refine
  5. link added 2d ago
    agent · flower-refine
  6. status change 2d ago
    agent · flower-refine
  7. dependency added 2d ago

    Now depends on #36 (Any reason we couldn't set up the brief view/detail page to allow a di).

    agent · flower-refine
  8. plan proposed 2d ago

    ## Goal Verify **direct-dispatch v1 (#36)** works end-to-end on real Solo agents before v2 (#104) is built — the operator-requested gate between v1 and v2. ## Depends on #36 (v1 must ship first) · Blocks #104 (v2 stays blocked until this passes) ## What to verify (v1 acceptance, exercised live — not just the faked test seam) 1. Brief-detail "Dispatch direct" lists a project's daemons grouped by worktree with correct liveness; only idle agents are selectable. 2. Selecting an idle agent + dispatching **pushes the brief packet into that agent's Solo process via solo-cli** and the agent actually receives + acts on it (the core v1 mechanism — verify the packet lands and wakes the idle agent). 3. A `direct` dispatch_request is recorded on the brief trace + dispatch history. 4. When an active orchestrator exists it receives the heads-up inbox message; when none exists (the motivating case) dispatch still works with no orchestrator. 5. Failure/edge behavior: agent not actually idle, solo-cli unreachable / `stale_discovery`, bad process id. ## Deliverable A pass/fail verification report on this brief covering the above, run on a REAL Solo agent. If issues are found, file them (feedback/bug) before v2 proceeds. ## Acceptance - Documented end-to-end verification of v1 on a live agent; an explicit pass so #104 can unblock. - `Brief: #103` trailer. ## Provenance Operator-requested middle verification gate (#36 answer, 2026-07-01). Depends on #36; blocks #104. Created + chained by flower-refine (2026-07-03).

    agent · flower-refine
  9. note added 2d ago

    Chained gate between #36 (direct-dispatch v1: direct-to-existing-idle-agent via solo-cli) and its v2. Verify v1 actually works end-to-end against real Solo agents before v2 is built. Blocked on #36. Per operator's #36 answer (2026-07-01): a middle brief to verify v1 works properly before v2 can be chained after it. Full spec via brief_update_spec.

    agent · flower-refine
  10. participant joined 2d ago
    system · flower-refine

epic · dependencies

Relationships

epic parent

depends on

agents · waves

Participants

  • flower-refine participant · active
  • flower-36 participant · active

trace · graph

Links

  • Scratchpad #378 execution
  • Scratchpad #375 execution
  • Scratchpad #364 execution

scope

Projects

  • flower · primary

dogfood · read-only

Agent’s-eye view

The literal recall_brief payload an agent gets — same service path as the MCP tool.