[rwasayc]

Bring the owned Android fork forward with its complete existing feature dependency chain, rather than publishing only the final testing documentation. The branch includes the recorded upstream sync through ade4730, retained Android SSH/localization and Mermaid customizations, the project-control modal with an explicitly unavailable host adapter, and per-thread read-only Harbor MCP binding for Codex.

Harbor configuration is thread/cwd/launcher-digest bound and does not install global MCP configuration or credentials. The testing adaptation adds TESTING.md, compatible explicit unit/component/browser/native profiles, required-suite nonempty behavior and consistent script-component CI selection. Existing broad CI and public command names are retained. Exact project-testing 1.0.0 was installed and verified locally; local managed skill/profile/state files are excluded.

Base: this owned fork's workbench/android-ssh branch. The 1048-commit range includes upstream synchronization and historical Android prerequisites; it is intentionally not an upstream pingdotgg PR. All custom-history blobs were scanned for credential patterns/large artifacts; detected key markers are short synthetic fixtures. Vendored certificate-extension rejection fixtures match the imported upstream tree. No user credentials, private runtime state, APKs or local installation/control files are included.

Validation: static Python AST, JSON/profile discovery and required-file checks; documented exact installation integrity, offline doctor and committed history. Earlier feature results remain bound to their original SHAs. No new combined check:changed/type/unit/component/browser/device/model run was performed for this head; sparse checkout has absent relay/native inputs. Remote Actions may run. Needs serialized quick profiles and affected UI/Harbor regressions before merge; no product/deployment/release or complete mobile-runtime PASS is claimed.

Prepared and integrated with Codex gpt-6.1-sol (high), T3 Code Codex harness; prior Android/UI implementation retains its recorded provenance.

Hosted CI follow-up: push and PR Testing profiles jobs script-components and core-type-and-unit both completed SUCCESS at 2993fd7. Scope: static entry contracts/inventory, registered six script components, original contracts/shared Effect typecheck and four serial unit files. Prior checkout/partial-clone failures remain historical failures. No new local heavy/model/UI run or merge.
Other gates remain: context budget rejects upstream-inherited +221/+221/+2007 rendered bytes versus actual fork base; no approval label or budget change made. PR-size ENOBUFS is fixed in separate prerequisite #2, which must be merged into the base before the base-owned target workflow can use it. Native/custom-runner checks can remain queued independently.

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ No successful main baseline artifact is available yet. This run establishes the initial measurement.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	—	5.0 KiB	—	6.8 KiB	✅
Codex	Thread snapshot wire	—	3.8 KiB	—	4.9 KiB	✅
Codex	Live turn WebSocket wire	—	1.2 KiB	—	2.0 KiB	✅
Codex	Live turn WebSocket decoded	—	21.5 KiB	—	24.4 KiB	✅
Codex	Live turn messages	—	1	—	8	✅
Claude	Total thread wire	—	5.0 KiB	—	6.8 KiB	✅
Claude	Thread snapshot wire	—	3.8 KiB	—	4.9 KiB	✅
Claude	Live turn WebSocket wire	—	1.2 KiB	—	2.0 KiB	✅
Claude	Live turn WebSocket decoded	—	21.9 KiB	—	24.4 KiB	✅
Claude	Live turn messages	—	1	—	8	✅

Baseline: unavailable · PR result: 6691867 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.