Repository navigation
Execution — reproduce one local agent task through Graph #8
Description
Outcome
One useful local task reaches the Agent, exact capability/grant, Graph and retained result under one named supported profile.
Next planning checkpoint: Init. This is the earliest scope/review conversation, not an assignment or a promise to finish the whole outcome in that week. Selected work starts at #52.
Remaining acceptance
This outcome consolidates the remaining selected-profile acceptance from #5, #6, #7 and #87. Reuse existing foundations; do not rebuild AgentForge, dispatch or orchestration because an old checkbox was unchecked.
- Select one exact task/model/profile; reproduce typed output plus bounded errors/cancellation and scope cleanup.
- Demonstrate eligibility/grant/lease handling and safe runtime activation only to the extent the selected profile actually supports.
- Retain the exact workflow, request and result across the required supported persistence boundary; distinguish model completion from a useful answer.
- Include the qualified model-local request policy from Admit a model-local non-thinking request policy for qualified Qwen chat #87 where that selected model needs it, without changing other runtimes.
- Record current source, actual test/host/model evidence and explicit remaining limits. Broader scheduling, policy and deployment qualification stay separately selected.
Evidence and handoff
Start from current State of Work and its linked source/tests and owning ADRs. The linked original acceptance remains binding unless a named replacement outcome records its disposition. Detailed scope, negative cases and dated campaign evidence are preserved in preserved scope and evidence; historical evidence does not establish current qualification on its own. Any accepted requirement not selected for this profile needs an explicit route in #62 before this issue closes.
Choose one small implementation or verification slice when this outcome is admitted. Record the responsible contributor, exact profile/revision, observed result, failure/unknown boundary and next checkpoint. The founder reviews and approves all code for this campaign; merge and promotion follow CONTRIBUTING.
Activity
[-]Graph — workflow execution, waits & recovery[/-][+]Graph — Basic: one local conversation workflow[/+][-]Graph — Basic: one local conversation workflow[/-][+]Graph — one local conversation workflow[/+]Fix Graph wait integrity and expired peer recovery
zenithruneblade commentedon Nov 6, 2026
Planning reset — 6 November 2026. The original acceptance below remains binding unless a named replacement outcome records its disposition. This is a preserved historical body, not a fresh capability/qualification claim.
Original issue scope and dated evidence before the planning reset
Execute one registered Bridge workflow through Agents, Dispatch and Orchestration without a browser.
Starting point: GraphRunner, workflow tests; delivery boundary.
Checklist
Boundary: retain existing supported recovery contracts. Hardware waits are not generally restart-durable; new wait types and parallel execution are later work.
Completion prerequisites: #5–#7 and the working startup portion of #18.
Suggested next step: #11 first browser conversation, within #9.
Later integration: #19 — A2A durable remote task and Graph resume; not required for this local conversation slice.
References: Graph (ADR 24) · Workers (ADR 14) · Workflow (ADR 28).
Follow-up: #22 durable service waits, #26 consent recovery and #29 registered Scrolls; #61 owns advanced nested subgraphs and parallel execution after the runtime-policy foundations. Execution adapters live in #27/#28; correlated evidence in #36.
Contribution entry: Follow one Run through the existing runner and fix one observed persistence/transition gap.
Verification: Graph tests and PostgreSQL; real local workflow for closure. Shared contribution and touched-docs rules: #52.
Delivery review — 12 October 2026
Published source: a913bff.
The serial Graph/worker path executes exact registered revisions with typed checkpoints, supported wait/re-entry and terminal recovery tests. Registry assembly now freezes private legacy_inline bindings and refuses drift before execution; public v2 snapshots remain unchanged. A real factory/PostgreSQL/SAQ/HTTP two-boot receipt passes with controlled collaborators, but the required production-persistence conversation through a real local model is still open.
Delivery evidence · Source/test 1 · Source/test 2
[-]Graph — one local conversation workflow[/-][+]Execution — reproduce one local agent task through Graph[/+]