From 28c2939c57e277c30294ce779b8b35b232d3623e Mon Sep 17 00:00:00 2001 From: Rafael Bittencourt Date: Tue, 22 Sep 2026 21:04:19 -0300 Subject: [PATCH] chore: rename GPT-5.6 Sol and Luna to the GPT-6 family Replace the retired gpt-5.6-sol and gpt-5.6-luna model ids with gpt-6-sol and gpt-6-luna in every lead config.toml and role card of the Advanced, Balanced and Lean arrangements, and in the README model tables. Allocation, reasoning effort and thread caps are unchanged: every Luna role stays at max, Sol keeps its effort, and Astra roles are untouched. Bump the three arrangements to 0.5.0 and record the change in the changelog, configuration and validation guides. The dated runtime evidence keeps the retired ids it was recorded against. --- README.md | 8 ++--- arrangements/advanced-delivery/README.md | 30 +++++++++++-------- .../agents/backend-worker.toml | 2 +- .../agents/code-reviewer.toml | 2 +- .../agents/database-engineer.toml | 2 +- .../advanced-delivery/agents/debugger.toml | 2 +- .../advanced-delivery/agents/explorer.toml | 2 +- .../agents/frontend-worker.toml | 2 +- .../advanced-delivery/agents/infra-sre.toml | 2 +- .../agents/integrator-reviewer.toml | 2 +- .../advanced-delivery/agents/researcher.toml | 2 +- .../agents/test-engineer.toml | 2 +- .../advanced-delivery/agents/test-runner.toml | 2 +- .../advanced-delivery/arrangement.toml | 2 +- arrangements/advanced-delivery/config.toml | 2 +- arrangements/balanced-delivery/README.md | 28 ++++++++++------- .../agents/backend-worker.toml | 2 +- .../agents/code-reviewer.toml | 2 +- .../agents/critical-reviewer.toml | 2 +- .../agents/database-engineer.toml | 2 +- .../balanced-delivery/agents/explorer.toml | 2 +- .../agents/frontend-worker.toml | 2 +- .../balanced-delivery/agents/infra-sre.toml | 2 +- .../agents/integrator-reviewer.toml | 2 +- .../balanced-delivery/agents/researcher.toml | 2 +- .../balanced-delivery/agents/test-runner.toml | 2 +- .../balanced-delivery/arrangement.toml | 2 +- arrangements/balanced-delivery/config.toml | 2 +- arrangements/lean-delivery/README.md | 22 +++++++++----- .../lean-delivery/agents/backend-worker.toml | 2 +- .../lean-delivery/agents/code-reviewer.toml | 2 +- .../agents/critical-reviewer.toml | 2 +- .../lean-delivery/agents/explorer.toml | 2 +- .../lean-delivery/agents/frontend-worker.toml | 2 +- .../agents/integrator-reviewer.toml | 2 +- .../lean-delivery/agents/researcher.toml | 2 +- arrangements/lean-delivery/arrangement.toml | 2 +- arrangements/lean-delivery/config.toml | 2 +- docs/CHANGELOG.md | 10 +++++++ docs/configuration.md | 3 +- docs/validation.md | 5 ++++ 41 files changed, 104 insertions(+), 70 deletions(-) diff --git a/README.md b/README.md index 85e710d..d663469 100644 --- a/README.md +++ b/README.md @@ -11,16 +11,16 @@ third-party orchestration service to use these files. [Skills](skills/README.md) · [Plugins](plugins/README.md) · [Validation and compatibility](docs/validation.md) -Version 0.4.0 is a community preview. Historical hook and selected-role tests are +Version 0.5.0 is a community preview. Historical hook and selected-role tests are recorded; the revised allocations and caps are not fully runtime-certified. ## Choose an arrangement | Arrangement | Lead | Specialist allocation | Configured cap / recommended active children | | --- | --- | --- | --- | -| [Advanced Delivery](arrangements/advanced-delivery/README.md) | `gpt-5.6-sol`, `xhigh` | 15 roles: 4 Astra, 4 Sol, 7 Luna | 4 / up to 3 | -| [Balanced Delivery](arrangements/balanced-delivery/README.md) | `gpt-5.6-sol`, `xhigh` | 12 roles: 2 Astra, 2 Sol, 8 Luna | 4 / up to 3 | -| [Lean Delivery](arrangements/lean-delivery/README.md) | `gpt-5.6-sol`, `high` | 8 roles: 1 Astra, 1 Sol, 6 Luna | 3 / up to 2 | +| [Advanced Delivery](arrangements/advanced-delivery/README.md) | `gpt-6-sol`, `xhigh` | 15 roles: 4 Astra, 4 Sol, 7 Luna | 4 / up to 3 | +| [Balanced Delivery](arrangements/balanced-delivery/README.md) | `gpt-6-sol`, `xhigh` | 12 roles: 2 Astra, 2 Sol, 8 Luna | 4 / up to 3 | +| [Lean Delivery](arrangements/lean-delivery/README.md) | `gpt-6-sol`, `high` | 8 roles: 1 Astra, 1 Sol, 6 Luna | 3 / up to 2 | The three arrangements provide progressively smaller role catalogs and enable Multi-Agent V2. Roles are available choices, not agents started together. Allocate only useful, diff --git a/arrangements/advanced-delivery/README.md b/arrangements/advanced-delivery/README.md index 37a050b..4ac9c64 100644 --- a/arrangements/advanced-delivery/README.md +++ b/arrangements/advanced-delivery/README.md @@ -1,6 +1,6 @@ # Advanced Delivery -Complex product delivery with dedicated reasoning for decisions, diagnosis, ordinary review, critical review, and coupled implementation. The lead uses `gpt-5.6-sol` at `xhigh`. This is an allocation policy, not a measured quality or cost guarantee. +Complex product delivery with dedicated reasoning for decisions, diagnosis, ordinary review, critical review, and coupled implementation. The lead uses `gpt-6-sol` at `xhigh`. This is an allocation policy, not a measured quality or cost guarantee. ## Install @@ -15,20 +15,20 @@ Reserve Sol for backend implementation, unknown-cause diagnosis, ordinary review | Role | Model | Reasoning effort | | --- | --- | --- | | advisor | gpt-6-astra | high | -| backend-worker | gpt-5.6-sol | high | -| code-reviewer | gpt-5.6-sol | high | +| backend-worker | gpt-6-sol | high | +| code-reviewer | gpt-6-sol | high | | critical-reviewer | gpt-6-astra | high | -| database-engineer | gpt-5.6-luna | max | -| debugger | gpt-5.6-sol | high | +| database-engineer | gpt-6-luna | max | +| debugger | gpt-6-sol | high | | design-lead | gpt-6-astra | high | -| explorer | gpt-5.6-luna | max | -| frontend-worker | gpt-5.6-luna | max | +| explorer | gpt-6-luna | max | +| frontend-worker | gpt-6-luna | max | | hard-task-specialist | gpt-6-astra | high | -| infra-sre | gpt-5.6-luna | max | -| integrator-reviewer | gpt-5.6-sol | high | -| researcher | gpt-5.6-luna | max | -| test-engineer | gpt-5.6-luna | max | -| test-runner | gpt-5.6-luna | max | +| infra-sre | gpt-6-luna | max | +| integrator-reviewer | gpt-6-sol | high | +| researcher | gpt-6-luna | max | +| test-engineer | gpt-6-luna | max | +| test-runner | gpt-6-luna | max | The parent supplies a complete bounded brief and inspects the delivered evidence. Every spawn uses `agent_type` and `fork_turns: "none"`, with no call-level model or effort override. Children never spawn children. @@ -44,6 +44,12 @@ This is an adapted public roster, not a full personal-profile export. Retained r ## Validation status +Revision 0.5.0 replaces the retired GPT-5.6 Sol and Luna model ids with +`gpt-6-sol` and `gpt-6-luna` in the lead and every role card. Allocation, +reasoning effort and caps are unchanged, and every Luna role stays at `max`. +This is a configuration change confirmed only by the catalog validators; it +has not been exercised in a new native run. + Revision 0.4.0 lowers this arrangement's total-thread cap from five to four (three active children, equal to Balanced Delivery) and rewrites the orchestration skill and both AGENTS snippets around bounded, single-mission delegation: an diff --git a/arrangements/advanced-delivery/agents/backend-worker.toml b/arrangements/advanced-delivery/agents/backend-worker.toml index 368686d..301f267 100644 --- a/arrangements/advanced-delivery/agents/backend-worker.toml +++ b/arrangements/advanced-delivery/agents/backend-worker.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Guido van Rossum Adapter", ] description = "Implement bounded backend work with clear acceptance and established patterns: APIs, services, jobs, business rules, integrations, and payment behavior. Deliver the diff and its lowest-layer validation. Also handles fully specified small edits; coupled invariants go to hard-task-specialist." -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "high" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/code-reviewer.toml b/arrangements/advanced-delivery/agents/code-reviewer.toml index cb77bbd..4009080 100644 --- a/arrangements/advanced-delivery/agents/code-reviewer.toml +++ b/arrangements/advanced-delivery/agents/code-reviewer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Bjarne Stroustrup Review", ] description = "Review an ordinary bounded diff for correctness, regressions, error handling, and missing tests. Read-only; deliver findings with file and line evidence, never a certification of your own implementation. Authorization, money, destructive migrations, or core invariants go to critical-reviewer." -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "high" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/database-engineer.toml b/arrangements/advanced-delivery/agents/database-engineer.toml index 756f036..f8f20fa 100644 --- a/arrangements/advanced-delivery/agents/database-engineer.toml +++ b/arrangements/advanced-delivery/agents/database-engineer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Peter Chen Entity Map", ] description = "Implement bounded database work: schema, queries, indexes, and migrations with explicit compatibility criteria, and decide the schema. Deliver the diff with forward and rollback validation. Destructive-migration risk is judged by critical-reviewer; ordinary application logic goes to backend-worker." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/debugger.toml b/arrangements/advanced-delivery/agents/debugger.toml index afd685d..2df58d1 100644 --- a/arrangements/advanced-delivery/agents/debugger.toml +++ b/arrangements/advanced-delivery/agents/debugger.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Rob Pike Repro Path", ] description = "Diagnose and fix ambiguous, intermittent, or cross-layer failures in the browser as well as the server. Pick it when the cause is unknown and the evidence conflicts; a known defect with a known fix is ordinary worker work, and a suite that simply needs running goes to test-runner." -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "high" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/explorer.toml b/arrangements/advanced-delivery/agents/explorer.toml index 3e3ad7b..412b1e1 100644 --- a/arrangements/advanced-delivery/agents/explorer.toml +++ b/arrangements/advanced-delivery/agents/explorer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Karen Sparck Jones Search Lens", ] description = "Locate entry points, callers, configuration, and tests within a bounded scope. Read-only; deliver file references and evidence. If the request needs writes, state the mismatch in one line and stop. Ambiguous diagnosis goes to debugger; an outside-source question goes to researcher." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/frontend-worker.toml b/arrangements/advanced-delivery/agents/frontend-worker.toml index 02a6aea..06b8218 100644 --- a/arrangements/advanced-delivery/agents/frontend-worker.toml +++ b/arrangements/advanced-delivery/agents/frontend-worker.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Nicole Sullivan Component Rule", ] description = "Implement bounded frontend work with established components, design, and state patterns. Deliver the diff with the rendered states checked. Interacting states, asynchronous flows, or performance constraints go to hard-task-specialist; unresolved design returns to the lead." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/infra-sre.toml b/arrangements/advanced-delivery/agents/infra-sre.toml index b9f711c..2d28466 100644 --- a/arrangements/advanced-delivery/agents/infra-sre.toml +++ b/arrangements/advanced-delivery/agents/infra-sre.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Niall Murphy Reliability Path", ] description = "Implement bounded CI, container, deployment-preparation, runtime, or reliability work with explicit blast radius and rollback, and assess release readiness. Deliver the diff and its validation; deployment execution remains outside the child scope. An unclear failure mechanism goes to debugger." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/integrator-reviewer.toml b/arrangements/advanced-delivery/agents/integrator-reviewer.toml index c41f416..384b42a 100644 --- a/arrangements/advanced-delivery/agents/integrator-reviewer.toml +++ b/arrangements/advanced-delivery/agents/integrator-reviewer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Rebecca Parsons System Fit", ] description = "Reconcile a bounded batch of worker deliverables against one acceptance contract. Read-only: read every touched file, run the scoped checks, and report integration defects for the lead to fix; deliver one consolidated report. High-risk certification goes to critical-reviewer." -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "high" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/researcher.toml b/arrangements/advanced-delivery/agents/researcher.toml index 7c16fc7..927cf97 100644 --- a/arrangements/advanced-delivery/agents/researcher.toml +++ b/arrangements/advanced-delivery/agents/researcher.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Eugene Garfield Citation Map", ] description = "Answer a question against primary sources outside this repository: official docs, specifications, changelogs, and installed dependency source. Read-only; every perishable fact comes back dated and cited or marked UNVERIFIED. Repository questions go to explorer instead." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/test-engineer.toml b/arrangements/advanced-delivery/agents/test-engineer.toml index 3760a6e..37f8c76 100644 --- a/arrangements/advanced-delivery/agents/test-engineer.toml +++ b/arrangements/advanced-delivery/agents/test-engineer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Elisabeth Hendrickson Acceptance Path", ] description = "Create the smallest deterministic reproduction and targeted regression coverage for a named behavior or bug, including bounded browser journeys. Deliver the tests and their run evidence. Implementation goes to the responsible worker; independent risk review goes to a reviewer." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/agents/test-runner.toml b/arrangements/advanced-delivery/agents/test-runner.toml index 65f4dcf..8009249 100644 --- a/arrangements/advanced-delivery/agents/test-runner.toml +++ b/arrangements/advanced-delivery/agents/test-runner.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Tibo Green Gate", ] description = "Run a named suite, read the output, and report what failed and why, with the smallest reproduction. Executes and diagnoses the log; it does not redesign the test architecture. A failure whose cause is genuinely unknown goes to debugger; new coverage goes to test-engineer." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/advanced-delivery/arrangement.toml b/arrangements/advanced-delivery/arrangement.toml index af06463..a8ca7e0 100644 --- a/arrangements/advanced-delivery/arrangement.toml +++ b/arrangements/advanced-delivery/arrangement.toml @@ -1,6 +1,6 @@ schema_version = 1 id = "advanced-delivery" -version = "0.4.0" +version = "0.5.0" author = "Rafael Bittencourt" description = "Advanced Delivery: Complex product delivery with dedicated reasoning for decisions, diagnosis, ordinary review, critical review, and coupled implementation." license = "MIT" diff --git a/arrangements/advanced-delivery/config.toml b/arrangements/advanced-delivery/config.toml index 38c4b1c..2e94f89 100644 --- a/arrangements/advanced-delivery/config.toml +++ b/arrangements/advanced-delivery/config.toml @@ -1,4 +1,4 @@ -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "xhigh" [features] diff --git a/arrangements/balanced-delivery/README.md b/arrangements/balanced-delivery/README.md index f6e8c50..17dfa66 100644 --- a/arrangements/balanced-delivery/README.md +++ b/arrangements/balanced-delivery/README.md @@ -1,6 +1,6 @@ # Balanced Delivery -Everyday product delivery with Luna implementing established patterns, Sol handling backend work and integration review, and Astra handling decisions and coupled implementation. The lead uses `gpt-5.6-sol` at `xhigh`. This is an allocation policy, not a measured quality or cost guarantee. +Everyday product delivery with Luna implementing established patterns, Sol handling backend work and integration review, and Astra handling decisions and coupled implementation. The lead uses `gpt-6-sol` at `xhigh`. This is an allocation policy, not a measured quality or cost guarantee. ## Install @@ -15,17 +15,17 @@ Reserve Astra for `advisor` and `hard-task-specialist`, and Sol for `backend-wor | Role | Model | Reasoning effort | | --- | --- | --- | | advisor | gpt-6-astra | high | -| backend-worker | gpt-5.6-sol | high | -| code-reviewer | gpt-5.6-luna | max | -| critical-reviewer | gpt-5.6-luna | max | -| database-engineer | gpt-5.6-luna | max | -| explorer | gpt-5.6-luna | max | -| frontend-worker | gpt-5.6-luna | max | +| backend-worker | gpt-6-sol | high | +| code-reviewer | gpt-6-luna | max | +| critical-reviewer | gpt-6-luna | max | +| database-engineer | gpt-6-luna | max | +| explorer | gpt-6-luna | max | +| frontend-worker | gpt-6-luna | max | | hard-task-specialist | gpt-6-astra | high | -| infra-sre | gpt-5.6-luna | max | -| integrator-reviewer | gpt-5.6-sol | high | -| researcher | gpt-5.6-luna | max | -| test-runner | gpt-5.6-luna | max | +| infra-sre | gpt-6-luna | max | +| integrator-reviewer | gpt-6-sol | high | +| researcher | gpt-6-luna | max | +| test-runner | gpt-6-luna | max | The parent supplies a complete bounded brief and inspects the delivered evidence. Every spawn uses `agent_type` and `fork_turns: "none"`, with no call-level model or effort override. Children never spawn children. @@ -41,6 +41,12 @@ This is an adapted public roster, not a full personal-profile export. Retained r ## Validation status +Revision 0.5.0 replaces the retired GPT-5.6 Sol and Luna model ids with +`gpt-6-sol` and `gpt-6-luna` in the lead and every role card. Allocation, +reasoning effort and caps are unchanged, and every Luna role stays at `max`. +This is a configuration change confirmed only by the catalog validators; it +has not been exercised in a new native run. + Revision 0.4.0 rewrites the orchestration skill and both AGENTS snippets around bounded, single-mission delegation: an accepted delivery closes the child's mission, a finished child is not reused for new work, and status or diff --git a/arrangements/balanced-delivery/agents/backend-worker.toml b/arrangements/balanced-delivery/agents/backend-worker.toml index 368686d..301f267 100644 --- a/arrangements/balanced-delivery/agents/backend-worker.toml +++ b/arrangements/balanced-delivery/agents/backend-worker.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Guido van Rossum Adapter", ] description = "Implement bounded backend work with clear acceptance and established patterns: APIs, services, jobs, business rules, integrations, and payment behavior. Deliver the diff and its lowest-layer validation. Also handles fully specified small edits; coupled invariants go to hard-task-specialist." -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "high" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/code-reviewer.toml b/arrangements/balanced-delivery/agents/code-reviewer.toml index 9115656..b3f3710 100644 --- a/arrangements/balanced-delivery/agents/code-reviewer.toml +++ b/arrangements/balanced-delivery/agents/code-reviewer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Bjarne Stroustrup Review", ] description = "Review an ordinary bounded diff for correctness, regressions, error handling, and missing tests. Read-only; deliver findings with file and line evidence, never a certification of your own implementation. Authorization, money, destructive migrations, or core invariants go to critical-reviewer." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/critical-reviewer.toml b/arrangements/balanced-delivery/agents/critical-reviewer.toml index caf94d6..a9472fa 100644 --- a/arrangements/balanced-delivery/agents/critical-reviewer.toml +++ b/arrangements/balanced-delivery/agents/critical-reviewer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Mary Shaw Assurance Check", ] description = "Independently review high-risk changes: authorization, billing, public APIs, destructive migrations, or core invariants. Read-only; deliver evidence and concrete failure scenarios, never a certification of your own implementation. Ordinary bounded diffs go to code-reviewer." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/database-engineer.toml b/arrangements/balanced-delivery/agents/database-engineer.toml index 756f036..f8f20fa 100644 --- a/arrangements/balanced-delivery/agents/database-engineer.toml +++ b/arrangements/balanced-delivery/agents/database-engineer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Peter Chen Entity Map", ] description = "Implement bounded database work: schema, queries, indexes, and migrations with explicit compatibility criteria, and decide the schema. Deliver the diff with forward and rollback validation. Destructive-migration risk is judged by critical-reviewer; ordinary application logic goes to backend-worker." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/explorer.toml b/arrangements/balanced-delivery/agents/explorer.toml index 096c95e..88894e4 100644 --- a/arrangements/balanced-delivery/agents/explorer.toml +++ b/arrangements/balanced-delivery/agents/explorer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Karen Sparck Jones Search Lens", ] description = "Locate entry points, callers, configuration, and tests within a bounded scope. Read-only; deliver file references and evidence. If the request needs writes, state the mismatch in one line and stop. Ambiguous diagnosis goes to hard-task-specialist; an outside-source question goes to researcher." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/frontend-worker.toml b/arrangements/balanced-delivery/agents/frontend-worker.toml index 02a6aea..06b8218 100644 --- a/arrangements/balanced-delivery/agents/frontend-worker.toml +++ b/arrangements/balanced-delivery/agents/frontend-worker.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Nicole Sullivan Component Rule", ] description = "Implement bounded frontend work with established components, design, and state patterns. Deliver the diff with the rendered states checked. Interacting states, asynchronous flows, or performance constraints go to hard-task-specialist; unresolved design returns to the lead." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/infra-sre.toml b/arrangements/balanced-delivery/agents/infra-sre.toml index 0254c40..47b2602 100644 --- a/arrangements/balanced-delivery/agents/infra-sre.toml +++ b/arrangements/balanced-delivery/agents/infra-sre.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Niall Murphy Reliability Path", ] description = "Implement bounded CI, container, deployment-preparation, runtime, or reliability work with explicit blast radius and rollback, and assess release readiness. Deliver the diff and its validation; deployment execution remains outside the child scope. An unclear failure mechanism goes to hard-task-specialist." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/integrator-reviewer.toml b/arrangements/balanced-delivery/agents/integrator-reviewer.toml index c41f416..384b42a 100644 --- a/arrangements/balanced-delivery/agents/integrator-reviewer.toml +++ b/arrangements/balanced-delivery/agents/integrator-reviewer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Rebecca Parsons System Fit", ] description = "Reconcile a bounded batch of worker deliverables against one acceptance contract. Read-only: read every touched file, run the scoped checks, and report integration defects for the lead to fix; deliver one consolidated report. High-risk certification goes to critical-reviewer." -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "high" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/researcher.toml b/arrangements/balanced-delivery/agents/researcher.toml index 7c16fc7..927cf97 100644 --- a/arrangements/balanced-delivery/agents/researcher.toml +++ b/arrangements/balanced-delivery/agents/researcher.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Eugene Garfield Citation Map", ] description = "Answer a question against primary sources outside this repository: official docs, specifications, changelogs, and installed dependency source. Read-only; every perishable fact comes back dated and cited or marked UNVERIFIED. Repository questions go to explorer instead." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/agents/test-runner.toml b/arrangements/balanced-delivery/agents/test-runner.toml index ff159a8..7410711 100644 --- a/arrangements/balanced-delivery/agents/test-runner.toml +++ b/arrangements/balanced-delivery/agents/test-runner.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Tibo Green Gate", ] description = "Run a named suite, read the output, and report what failed and why, with the smallest reproduction. Executes and diagnoses the log; it does not redesign the test architecture. A failure whose cause is genuinely unknown goes to hard-task-specialist; new coverage goes to the owning implementation worker." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/balanced-delivery/arrangement.toml b/arrangements/balanced-delivery/arrangement.toml index aeda30e..26da572 100644 --- a/arrangements/balanced-delivery/arrangement.toml +++ b/arrangements/balanced-delivery/arrangement.toml @@ -1,6 +1,6 @@ schema_version = 1 id = "balanced-delivery" -version = "0.4.0" +version = "0.5.0" author = "Rafael Bittencourt" description = "Balanced Delivery: Everyday product delivery with Luna implementing established patterns, Sol handling backend work and integration review, and Astra handling decisions and coupled implementation." license = "MIT" diff --git a/arrangements/balanced-delivery/config.toml b/arrangements/balanced-delivery/config.toml index 38c4b1c..2e94f89 100644 --- a/arrangements/balanced-delivery/config.toml +++ b/arrangements/balanced-delivery/config.toml @@ -1,4 +1,4 @@ -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "xhigh" [features] diff --git a/arrangements/lean-delivery/README.md b/arrangements/lean-delivery/README.md index 2f59bab..6ca6ef6 100644 --- a/arrangements/lean-delivery/README.md +++ b/arrangements/lean-delivery/README.md @@ -1,6 +1,6 @@ # Lean Delivery -Cost-conscious, well-bounded delivery with Luna implementing and reviewing, Sol reconciling deliveries, and one Astra advisor for decisions. The lead uses `gpt-5.6-sol` at `high`. This is an allocation policy, not a measured quality or cost guarantee. +Cost-conscious, well-bounded delivery with Luna implementing and reviewing, Sol reconciling deliveries, and one Astra advisor for decisions. The lead uses `gpt-6-sol` at `high`. This is an allocation policy, not a measured quality or cost guarantee. ## Install @@ -15,13 +15,13 @@ Use at most one active `advisor` and one active `integrator-reviewer`; other chi | Role | Model | Reasoning effort | | --- | --- | --- | | advisor | gpt-6-astra | high | -| backend-worker | gpt-5.6-luna | max | -| code-reviewer | gpt-5.6-luna | max | -| critical-reviewer | gpt-5.6-luna | max | -| explorer | gpt-5.6-luna | max | -| frontend-worker | gpt-5.6-luna | max | -| integrator-reviewer | gpt-5.6-sol | high | -| researcher | gpt-5.6-luna | max | +| backend-worker | gpt-6-luna | max | +| code-reviewer | gpt-6-luna | max | +| critical-reviewer | gpt-6-luna | max | +| explorer | gpt-6-luna | max | +| frontend-worker | gpt-6-luna | max | +| integrator-reviewer | gpt-6-sol | high | +| researcher | gpt-6-luna | max | The parent supplies a complete bounded brief and inspects the delivered evidence. Every spawn uses `agent_type` and `fork_turns: "none"`, with no call-level model or effort override. Children never spawn children. @@ -37,6 +37,12 @@ This is an adapted public roster, not a full personal-profile export. Retained r ## Validation status +Revision 0.5.0 replaces the retired GPT-5.6 Sol and Luna model ids with +`gpt-6-sol` and `gpt-6-luna` in the lead and every role card. Allocation, +reasoning effort and caps are unchanged, and every Luna role stays at `max`. +This is a configuration change confirmed only by the catalog validators; it +has not been exercised in a new native run. + Revision 0.4.0 rewrites the orchestration skill and both AGENTS snippets around bounded, single-mission delegation: an accepted delivery closes the child's mission, a finished child is not reused for new work, and status or diff --git a/arrangements/lean-delivery/agents/backend-worker.toml b/arrangements/lean-delivery/agents/backend-worker.toml index fd63174..f1363fc 100644 --- a/arrangements/lean-delivery/agents/backend-worker.toml +++ b/arrangements/lean-delivery/agents/backend-worker.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Guido van Rossum Adapter", ] description = "Implement bounded backend work with clear acceptance and established patterns: APIs, services, jobs, business rules, integrations, and payment behavior. Deliver the diff and its lowest-layer validation. Also handles fully specified small edits; coupled invariants go to the parent." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/lean-delivery/agents/code-reviewer.toml b/arrangements/lean-delivery/agents/code-reviewer.toml index 9115656..b3f3710 100644 --- a/arrangements/lean-delivery/agents/code-reviewer.toml +++ b/arrangements/lean-delivery/agents/code-reviewer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Bjarne Stroustrup Review", ] description = "Review an ordinary bounded diff for correctness, regressions, error handling, and missing tests. Read-only; deliver findings with file and line evidence, never a certification of your own implementation. Authorization, money, destructive migrations, or core invariants go to critical-reviewer." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/lean-delivery/agents/critical-reviewer.toml b/arrangements/lean-delivery/agents/critical-reviewer.toml index caf94d6..a9472fa 100644 --- a/arrangements/lean-delivery/agents/critical-reviewer.toml +++ b/arrangements/lean-delivery/agents/critical-reviewer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Mary Shaw Assurance Check", ] description = "Independently review high-risk changes: authorization, billing, public APIs, destructive migrations, or core invariants. Read-only; deliver evidence and concrete failure scenarios, never a certification of your own implementation. Ordinary bounded diffs go to code-reviewer." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/lean-delivery/agents/explorer.toml b/arrangements/lean-delivery/agents/explorer.toml index da284ce..417b45a 100644 --- a/arrangements/lean-delivery/agents/explorer.toml +++ b/arrangements/lean-delivery/agents/explorer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Karen Sparck Jones Search Lens", ] description = "Locate entry points, callers, configuration, and tests within a bounded scope. Read-only; deliver file references and evidence. If the request needs writes, state the mismatch in one line and stop. Ambiguous diagnosis goes to the parent; an outside-source question goes to researcher." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/lean-delivery/agents/frontend-worker.toml b/arrangements/lean-delivery/agents/frontend-worker.toml index 072bc83..c4fae80 100644 --- a/arrangements/lean-delivery/agents/frontend-worker.toml +++ b/arrangements/lean-delivery/agents/frontend-worker.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Nicole Sullivan Component Rule", ] description = "Implement bounded frontend work with established components, design, and state patterns. Deliver the diff with the rendered states checked. Interacting states, asynchronous flows, or performance constraints go to the parent; unresolved design returns to the lead." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/lean-delivery/agents/integrator-reviewer.toml b/arrangements/lean-delivery/agents/integrator-reviewer.toml index c41f416..384b42a 100644 --- a/arrangements/lean-delivery/agents/integrator-reviewer.toml +++ b/arrangements/lean-delivery/agents/integrator-reviewer.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Rebecca Parsons System Fit", ] description = "Reconcile a bounded batch of worker deliverables against one acceptance contract. Read-only: read every touched file, run the scoped checks, and report integration defects for the lead to fix; deliver one consolidated report. High-risk certification goes to critical-reviewer." -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "high" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/lean-delivery/agents/researcher.toml b/arrangements/lean-delivery/agents/researcher.toml index 7c16fc7..927cf97 100644 --- a/arrangements/lean-delivery/agents/researcher.toml +++ b/arrangements/lean-delivery/agents/researcher.toml @@ -8,7 +8,7 @@ nickname_candidates = [ "Eugene Garfield Citation Map", ] description = "Answer a question against primary sources outside this repository: official docs, specifications, changelogs, and installed dependency source. Read-only; every perishable fact comes back dated and cited or marked UNVERIFIED. Repository questions go to explorer instead." -model = "gpt-5.6-luna" +model = "gpt-6-luna" model_reasoning_effort = "max" developer_instructions = """ Return every result, blocker, and role recommendation to the parent. References to other roles are suggestions for the parent, never child-to-child handoffs. Do not contact or wait for another child; the parent decides whether to act, spawn, or resume a role. diff --git a/arrangements/lean-delivery/arrangement.toml b/arrangements/lean-delivery/arrangement.toml index 3c0ff8d..8f8daf4 100644 --- a/arrangements/lean-delivery/arrangement.toml +++ b/arrangements/lean-delivery/arrangement.toml @@ -1,6 +1,6 @@ schema_version = 1 id = "lean-delivery" -version = "0.4.0" +version = "0.5.0" author = "Rafael Bittencourt" description = "Lean Delivery: Cost-conscious, well-bounded delivery with Luna implementing and reviewing, Sol reconciling deliveries, and one Astra advisor for decisions." license = "MIT" diff --git a/arrangements/lean-delivery/config.toml b/arrangements/lean-delivery/config.toml index b142d24..9419109 100644 --- a/arrangements/lean-delivery/config.toml +++ b/arrangements/lean-delivery/config.toml @@ -1,4 +1,4 @@ -model = "gpt-5.6-sol" +model = "gpt-6-sol" model_reasoning_effort = "high" [features] diff --git a/docs/CHANGELOG.md b/docs/CHANGELOG.md index 63c992b..e0fc356 100644 --- a/docs/CHANGELOG.md +++ b/docs/CHANGELOG.md @@ -1,5 +1,15 @@ # Changelog +## 0.5.0 — 2026-09-22 + +- Replace the retired `gpt-5.6-sol` and `gpt-5.6-luna` model ids with + `gpt-6-sol` and `gpt-6-luna` in all three arrangements: lead `config.toml`, + every role card, and the README model tables. +- Allocation, reasoning effort and thread caps are unchanged; every Luna role + remains at `max` and Astra roles are untouched. +- Configuration change only; the dated runtime evidence still refers to the + retired ids and has not been re-run against the new models. + ## 0.4.0 — 2026-09-20 - Lower Advanced Delivery's total-thread cap from five to four, matching diff --git a/docs/configuration.md b/docs/configuration.md index fc6f266..4426acd 100644 --- a/docs/configuration.md +++ b/docs/configuration.md @@ -14,7 +14,7 @@ ## Native child wait settings -All three arrangements use the same wait settings in version 0.4.0: +All three arrangements use the same wait settings in version 0.5.0: | Field under `features.multi_agent_v2` | Native baseline | Catalog value | Meaning | | --- | --- | --- | --- | @@ -63,6 +63,7 @@ and [default nickname selection](https://github.com/openai/codex/blob/b979d4f1f0 | Version | Change | Reason | Rollback | | --- | --- | --- | --- | +| 0.5.0 (2026-09-22) | Replace `gpt-5.6-sol` and `gpt-5.6-luna` with `gpt-6-sol` and `gpt-6-luna` in every lead and role card; allocation, effort and caps unchanged | The GPT-5.6 Sol and Luna ids are retired | Restore the prior `model` values in each `config.toml` and `agents/*.toml`; no other field changed. | | 0.4.0 (2026-09-20) | Lower Advanced Delivery's total-thread cap from five to four (three active children, equal to Balanced); rewrite the orchestration skill and every AGENTS snippet around bounded, single-mission delegation | Treat the configured cap as a ceiling, not a target, and stop the context cost of reusing or repeatedly waking a child after its mission is accepted | Restore `max_concurrent_threads_per_session = 5` in Advanced Delivery's `config.toml`; restore the prior skill and snippet wording. No other arrangement's config changed. | | 0.3.0 (2026-09-13) | Set one-minute minimum and five-minute default child waits; add adaptive waiting guidance | Reduce empty parent wait returns without changing budgets or the execution topology | Restore only prior wait fields and the affected instruction/skill sections; remove these fields only if they were previously absent. | | 0.2.0 (2026-09-13) | Add Balanced Delivery; rebalance models, reduce catalogs to 15/12/8 roles and total-thread caps to 5/4/3; require parent-mediated routing | Match reasoning to responsibility and reduce selection and coordination overhead | Reinstall reviewed prior role cards and matching snippets/config fields together. Follow the upgrade removal list; preserve credentials and unrelated settings. | diff --git a/docs/validation.md b/docs/validation.md index 8d905da..7716293 100644 --- a/docs/validation.md +++ b/docs/validation.md @@ -75,6 +75,11 @@ with intended child limits of 4/3/2. Those revised values and role allocations have not been exercised in a new native saturation or complete role-coverage run. Do not treat the historical writer or hook runs as certification of the new matrix. +Revision 0.5.0 renames the Sol and Luna model ids to `gpt-6-sol` and `gpt-6-luna` +without changing allocation, effort or caps. The dated runtime evidence below +was recorded against the retired GPT-5.6 Sol and Luna ids and does +not certify the renamed models. + Revision 0.4.0 lowers Advanced from 5 to 4 total threads (3 children), equal to Balanced, on the stated ground that the cap is a ceiling and not a delegation target; Balanced and Lean's caps are unchanged. This is a configuration and