estafeta 3.2.0: provider policy, takeover guard, Windows, live checks across hosts

.github/workflows/ci.yml

@@ -2,6 +2,7 @@ name: CI
 
 on:
   push:
+    branches: [main]
   pull_request:
 
 permissions:
@@ -12,7 +13,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-latest, macos-latest]
+        os: [ubuntu-latest, macos-latest, windows-latest]
         python: ["3.9", "3.13"]
     runs-on: ${{ matrix.os }}
     timeout-minutes: 20
@@ -24,8 +25,11 @@ jobs:
       - name: Pack validation
         run: python scripts/workflow.py check
       - name: Runtime tests
+        if: success() || failure()
         run: python scripts/test_session.py
       - name: Installer tests
+        if: success() || failure()
         run: python scripts/test_workflow.py
       - name: Golden path integration tests (stub worker)
+        if: success() || failure()
         run: python scripts/test_relay.py

HANDOVER.md

@@ -2,74 +2,99 @@
 
 ## Current work
 
-estafeta 3.1.0 adds cross-provider work (plan:
-docs/plans/multi-provider.md).
-
-- **Skill `agent-relay`.** Lets an orchestrator consult, cross-review with or
-  delegate a phase to another provider, and coordinate peer agents. It covers
-  T3 host tasks, Codex/Claude/Gemini CLIs and same-provider subagents.
-- **Runtime command `delegate`.** Keeps a ledger beside the state, so schema 2
-  is unchanged. CLI workers run as supervised processes with a timeout. The
-  runtime checks a writer's claims against recorded checks and the observed
-  workspace.
-- **Integrated.** Existing skills, docs and the 3.1.0 manifests know about
-  the skill. A CI workflow was added.
-- **Testing approach (user request).** Golden path tests check a short
-  sequence of agent interactions, with an expected result at each step:
-  - every change, in CI: a deterministic stub worker
-    (`scripts/test_relay.py`);
-  - before a release or a provider/model change: live runs checked with
-    `--live-verify`.
+3.1.0 (`agent-relay` and runtime `delegate`) is merged to main through PR #3.
+
+Release 3.2.0 is in progress on branch `relay-hardening-3.1.1` (draft PR #5 to
+main; plan docs/plans/relay-hardening.md). Done so far:
+
+- **Windows.** Supported and verified in CI. Fixed: index file modes in
+  fingerprints, job-object supervision, and a file-based stop request for
+  nested checks.
+- **Takeover guard.** A solo session-start no longer takes over a live thread
+  in the same folder. This was observed live with a parallel Codex thread in
+  T3.
+- **Provider policy.** Projects get an "AI providers" table that session-start
+  asks for on first use and `delegate` enforces. This project allows every
+  available provider except the Claude (Pricefx) work account.
+- **Live checks.** These passed:
+  - `codex exec` as a worker;
+  - Antigravity through T3;
+  - orchestrators in Claude Code and in Codex, each with the other's CLI and
+    with their own native subagents;
+  - the Gemini CLI failure mode without an API key.
+- **Drivers.** `scripts/golden/live_cli.py` and `live_orchestrator.py` replay
+  the live golden paths.
+
+PR #4 was merged only into its stale base branch `agent-relay-3.1.0`, so its
+commits are not in main. PR #5 carries them.
 
 ## Next action
 
-Merge the 3.1.0 pull request after its CI run passes. Optional follow-ups:
+The work moves to another computer. There, follow docs/CONTINUING.md with
+branch `relay-hardening-3.1.1` instead of `main`, and use session-start. The
+runtime state (phases, checks, the delegation ledger) stays on this computer,
+so import the open phases R4 and R5 from the plan.
 
-- a third Codex review round;
-- live `codex exec` and `gemini -p` runs where those CLIs are installed;
-- the benchmark follow-up in docs/BENCHMARK-ROADMAP.md.
+Then fix the 11 findings of Codex review round 3. They are listed in
+docs/plans/relay-hardening.md, "Status at pause". Start with the
+high-severity ones:
+
+1. the policy parser;
+2. the Windows job assignment race;
+3. nested stop chaining;
+4. the late-recorded host writer baseline.
+
+Add a failing test for each, then rerun the suites and the affected live
+paths. Have Codex re-review the fix delta. Then release 3.2.0 (R5) and mark PR
+#5 ready.
 
 ## Validation
 
 On October 15, 2026, on macOS with Python 3.9.6:
 
-- runtime 70/70, installer 21/21, golden 5/5 scenarios (79 steps), pack check
-  and the Claude plugin validators passed;
-- four deliberate regressions were caught by golden steps;
-- live, all passing `--live-verify`:
-  - Codex gpt-6.1-sol through T3: a delegated write and two cross-review
-    rounds (12, then 5 findings, all fixed with tests);
-  - `claude -p` as a CLI worker: consult on Haiku 5.5, write on Sonnet 5.5
-    under the OS sandbox.
-
-Details are in packs/session-workflow/VALIDATION.md.
+- runtime 76/76, installer 21/21 and golden 6/6 scenarios passed, and
+  `workflow.py check` passed.
+- PR #5 CI was green on Linux, macOS and Windows at 29fb17e, the last code
+  change; this closed R1 on October 16.
+- Live evidence for each transport and host is in
+  packs/session-workflow/VALIDATION.md.
 
 ## Boundaries
 
-Not verified:
-
-- live `codex exec` and `gemini -p` CLI transports;
-- the Gemini/Antigravity route;
-- native Linux and Windows.
-
-The round 2 fixes were tested locally but not re-reviewed by Codex. The
-fixtures in /private/tmp/relay-live and /private/tmp/relay-cli are disposable.
+- **Unverified:**
+  - a live `gemini -p` answer (no API key);
+  - Codex orchestrating without network access (not run);
+  - native Windows beyond GitHub CI.
+- **Machine-local:** the runtime state and the live-run fixtures under the
+  temp directory stay on the first computer. VALIDATION.md keeps the live
+  evidence.
+- **Installs:** the global estafeta skills in the provider profiles are outdated
+  (2.4.1 or missing). Update them with `python3 scripts/workflow.py install
+  --all-detected` when convenient.
 
 <!-- session-workflow:facts:start -->
 ## Session facts (generated)
 
-State revision 109; HEAD 8559853e8949b0ef51d5ffb9c2351158fba23ff2.
+State revision 162; HEAD 29fb17eaebe23d48326a73ee436093ce841035ab.
 
 | Phase | State | Owner | Latest checks |
 | --- | --- | --- | --- |
-| P1 | DONE | - | unit-session: pass (executed, inputs match); golden-relay: pass (executed, inputs match); pack-check: pass (executed, inputs match) |
-| P2 | DONE | - | pack-check: pass (executed, inputs match); unit-workflow: pass (executed, inputs match) |
-| P3 | DONE | - | pack-check: pass (executed, inputs match); unit-workflow: pass (executed, inputs match); unit-session: pass (executed, inputs match); golden-relay: pass (executed, inputs match) |
-| P4 | DONE | - | live-relay: pass (executed, inputs match); unit-session: pass (executed, inputs match); golden-relay: pass (executed, inputs match) |
-| P5 | DONE | - | live-cli: pass (executed, inputs match); live-review-2: pass (executed, inputs match); golden-relay: pass (executed, inputs match); pack-check: pass (executed, inputs match) |
+| P1 | DONE | - | unit-session: pass (executed, inputs changed); golden-relay: pass (executed, inputs changed); pack-check: pass (executed, inputs changed) |
+| P2 | DONE | - | pack-check: pass (executed, inputs changed); unit-workflow: pass (executed, inputs changed) |
+| P3 | DONE | - | pack-check: pass (executed, inputs changed); unit-workflow: pass (executed, inputs changed); unit-session: pass (executed, inputs changed); golden-relay: pass (executed, inputs changed) |
+| P4 | DONE | - | live-relay: pass (executed, inputs changed); unit-session: pass (executed, inputs changed); golden-relay: pass (executed, inputs changed) |
+| P5 | DONE | - | live-cli: pass (executed, inputs changed); live-review-2: pass (executed, inputs changed); golden-relay: pass (executed, inputs changed); pack-check: pass (executed, inputs changed) |
+| R1 | DONE | - | unit-session: pass (executed, inputs match); unit-workflow: pass (executed, inputs match); golden-relay: pass (executed, inputs match); pack-check: pass (executed, inputs match); ci-windows: pass (executed, inputs match) |
+| R2 | DONE | - | live-codex-cli: pass (executed, inputs match) |
+| R3 | DONE | - | live-antigravity: pass (executed, inputs match); gemini-cli-failure: pass (executed, inputs match) |
+| R4 | TODO | - | - |
+| R5 | TODO | - | - |
+| R6 | DONE | - | unit-session: pass (executed, inputs changed); golden-relay: pass (executed, inputs changed); pack-check: pass (executed, inputs changed) |
+| R7 | DONE | - | unit-session: pass (executed, inputs match); golden-relay: pass (executed, inputs match) |
+| R8 | DONE | - | live-orchestrators: pass (executed, inputs match) |
 
 Active sessions: none.
-Ended sessions: 5; latest s-xprov-1015 (next: Watch CI of the agent-relay-3.1.0 pull request and merge it when green. Optional: Codex review round 3, live codex exec / gemini -p runs where installed, benchmark follow-up.).
+Ended sessions: 10; latest s-transfer-1016 (next: On another computer: follow docs/CONTINUING.md with branch relay-hardening-3.1.1, use session-start, import open phases R4 and R5 from docs/plans/relay-hardening.md, then fix the 11 Codex review round 3 findings (high severity first, each with a failing test), have Codex re-review the fix delta, release 3.2.0 (R5) and mark PR #5 ready).
 
 Generated facts do not establish acceptance, explain intent or prove delivery.
 <!-- session-workflow:facts:end -->

PROJECT.md

@@ -8,16 +8,40 @@ Do not publish credentials, local runtime state, or personal logs.
 
 ## Documentation map
 
-- Plan: docs/plans/multi-provider.md (multi-provider relay).
+- Plan: docs/plans/relay-hardening.md (3.1.1); previous: docs/plans/multi-provider.md.
 - Handover: HANDOVER.md.
 - Installation: docs/INSTALL.md; usage: docs/USAGE.md.
 - Maintenance and checks: docs/MAINTAINING.md.
 - Workflow: estafeta; machine state in .agent-state/session-workflow is local and ignored by Git.
 
 ## Current focus
 
-- Multi-provider relay (3.1.0): skill `agent-relay` and runtime `delegate`; see docs/plans/multi-provider.md.
-- On another computer, follow docs/CONTINUING.md, then use session-start.
+- 3.1.0 (multi-provider relay) is merged to main. Release 3.2.0 is in progress
+  on branch `relay-hardening-3.1.1`, draft PR #5 to main (plan:
+  docs/plans/relay-hardening.md). PR #4 was merged only into the stale base
+  branch `agent-relay-3.1.0`, not into main.
+- Paused 2026-10-15. Next: fix the 11 findings of Codex review round 3 (R4).
+- On another computer, follow docs/CONTINUING.md with branch
+  `relay-hardening-3.1.1` instead of `main`, then use session-start. The
+  runtime state stays on the first computer: import the open phases R4 and R5
+  from the plan there.
+
+## AI providers
+
+Decided by the user on 2026-10-15: every available provider except the
+Claude (Pricefx) work account. Agents ask before changing this table.
+Providers not listed (for example Cursor, Grok or OpenCode, currently
+disabled in T3) are refused until the user adds them.
+
+<!-- estafeta:providers:start -->
+| Provider | Allowed | Models | Notes |
+| --- | --- | --- | --- |
+| claude | orchestrate, consult, review, write | any | Personal account: T3 instance `claudeAgent`, Claude Code, `claude -p` |
+| claude-pricefx | none | - | Work account (T3 instance `claude-pricefx`): not for this project |
+| codex | orchestrate, consult, review, write | any | T3 tasks and `codex exec` (local login) |
+| antigravity | orchestrate, consult, review, write | any | T3 tasks; Gemini models without an API key |
+| gemini | consult, review, write | any | `gemini -p`; needs a paid API key, none on this host yet |
+<!-- estafeta:providers:end -->
 
 ## Run and verify
 

docs/USAGE.md

@@ -66,8 +66,14 @@ transfer it together with its evidence.
 
 - **Solo (default):** one writer per checkout. A new chat supersedes the earlier
   chats in the same checkout and takes over their phase claims and preview
-  servers. A chat closed without session-end therefore blocks nothing. A chat that
-  is still running a check is detected as alive, and the takeover is refused.
+  servers, so a chat closed without session-end blocks nothing for long.
+  - **No takeover of live chats.** A new chat does not take over a chat that
+    may still be working: one running a check, one that acted in the last 30
+    minutes, or one with open delegations. It joins in parallel and says so.
+    This matters in hosts such as T3 Code, where several threads share one
+    folder.
+  - **Idle threshold.** `SESSION_WORKFLOW_TAKEOVER_IDLE_MINUTES` changes the 30
+    minutes.
 - **Parallel:** tell a chat, in any language, that another agent works in the
   same folder ("work in parallel", «работаем параллельно», «pracujeme
   paralelně», «並行して作業する»), and it begins with `--parallel`. The agent goes