Record the completed Gate 0 baseline

Record the completed Gate 0 baseline

Deploy story

What this commit did to the running system — joined from the forge receipt chain, the part a commit page elsewhere cannot show.

Not deployed through the forge lane

No push, promotion, build, or deploy receipt references this commit (receipts are scanned over a bounded recent window). Changes shipped by full node replacement carry their proof in the release gate receipt instead.

Changed files

modified docs/2026-08-20-integration-hardening-and-staging-readiness-recommendations.md

Diff

1 file changed, +20 -7


Date: 2026-08-20

Status: Proposed, amended with measured findings
Status: In progress; Gate 0 complete, amended with measured findings

## Outcome


- Replaced Ra's captured session-query functions with stable
  module-function-argument descriptors. The cluster suite now runs under
  coverage without a peer crashing on an instrumented function identity.
- Merged the default and cluster coverage exports locally. The current combined
  result is 83.58% from 1,222 default tests and all 9 cluster tests.
- Merged the default and cluster coverage exports locally. The completed
  baseline result is 83.59% from 1,222 default tests and all 9 cluster tests.
- Added `ops/ci/coverage.sh` to discard stale exports, run both suites with
  warnings as errors, collect execution from distributed peers, merge both
  exports, and enforce an initial 83% floor. Raise the floor as direct recovery

  coverage, and the production release smoke without retries, recheck Git
  identity and cleanliness, and atomically write a content-free receipt under
  `.git/openagents/gate-receipts/`.

Gate 0 still requires one clean-commit execution of `ops/ci/baseline.sh` and
inspection of its exact-SHA receipt.
- Ran the complete baseline on clean commit
  `d9ffc65f5cdd961cf228146a95e9d651e14692d2`. It passed without automatic
  retries in 76 seconds: 15 JavaScript tests, 1,222 default Elixir tests, all 9
  separately executed cluster tests, 83.59% merged coverage, and the packaged
  production release smoke against a disposable PostgreSQL 18 database.
- Inspected the resulting mode-`0600`, content-free local receipt at
  `.git/openagents/gate-receipts/d9ffc65f5cdd961cf228146a95e9d651e14692d2.json`.
  The receipt identifies the exact commit, stage results and durations, bounded
  test counts, coverage, and zero automatic retries; it contains no logs,
  database URLs, hostnames, credentials, or product content.

Gate 0 is complete. Run `ops/ci/baseline.sh` again on every subsequent release
candidate; a receipt for one SHA is never evidence for another.

Do not use the current green suite as evidence for untested code. The updated
coverage audit records strong Issues and Projects coverage and the defects it

stable module-function-argument descriptors, and all 9 cluster tests complete
under coverage. Coverage-aware peer shutdown now collects remote execution, and
direct `RaBootstrap` decision tests cover the previously untested worker. The
merged default and cluster result is 83.58% and passes the enforced 83% floor.
completed exact-SHA baseline supersedes that intermediate result. At
`d9ffc65f5cdd961cf228146a95e9d651e14692d2`, the merged default and cluster
result is 83.59% and passes the enforced 83% floor. The release smoke also
passes, and the inspected local receipt records zero automatic retries.

## A2. Blocker: staging is not isolated from production today


2	2
3	3		`Date: 2026-08-20`
4	4
5		-	`Status: Proposed, amended with measured findings`
	5	+	`Status: In progress; Gate 0 complete, amended with measured findings`
6	6
7	7		`## Outcome`
8	8

94	94		`- Replaced Ra's captured session-query functions with stable`
95	95		`module-function-argument descriptors. The cluster suite now runs under`
96	96		`coverage without a peer crashing on an instrumented function identity.`
97		-	`- Merged the default and cluster coverage exports locally. The current combined`
98		-	`result is 83.58% from 1,222 default tests and all 9 cluster tests.`
	97	+	`- Merged the default and cluster coverage exports locally. The completed`
	98	+	`baseline result is 83.59% from 1,222 default tests and all 9 cluster tests.`
99	99		- Added `ops/ci/coverage.sh` to discard stale exports, run both suites with
100	100		`warnings as errors, collect execution from distributed peers, merge both`
101	101		`exports, and enforce an initial 83% floor. Raise the floor as direct recovery`

115	115		`coverage, and the production release smoke without retries, recheck Git`
116	116		`identity and cleanliness, and atomically write a content-free receipt under`
117	117		`.git/openagents/gate-receipts/`.
118		-
119		-	Gate 0 still requires one clean-commit execution of `ops/ci/baseline.sh` and
120		-	`inspection of its exact-SHA receipt.`
	118	+	`- Ran the complete baseline on clean commit`
	119	+	`d9ffc65f5cdd961cf228146a95e9d651e14692d2`. It passed without automatic
	120	+	`retries in 76 seconds: 15 JavaScript tests, 1,222 default Elixir tests, all 9`
	121	+	`separately executed cluster tests, 83.59% merged coverage, and the packaged`
	122	+	`production release smoke against a disposable PostgreSQL 18 database.`
	123	+	- Inspected the resulting mode-`0600`, content-free local receipt at
	124	+	`.git/openagents/gate-receipts/d9ffc65f5cdd961cf228146a95e9d651e14692d2.json`.
	125	+	`The receipt identifies the exact commit, stage results and durations, bounded`
	126	+	`test counts, coverage, and zero automatic retries; it contains no logs,`
	127	+	`database URLs, hostnames, credentials, or product content.`
	128	+
	129	+	Gate 0 is complete. Run `ops/ci/baseline.sh` again on every subsequent release
	130	+	`candidate; a receipt for one SHA is never evidence for another.`
121	131
122	132		`Do not use the current green suite as evidence for untested code. The updated`
123	133		`coverage audit records strong Issues and Projects coverage and the defects it`

1103	1113		`stable module-function-argument descriptors, and all 9 cluster tests complete`
1104	1114		`under coverage. Coverage-aware peer shutdown now collects remote execution, and`
1105	1115		direct `RaBootstrap` decision tests cover the previously untested worker. The
1106		-	`merged default and cluster result is 83.58% and passes the enforced 83% floor.`
	1116	+	`completed exact-SHA baseline supersedes that intermediate result. At`
	1117	+	`d9ffc65f5cdd961cf228146a95e9d651e14692d2`, the merged default and cluster
	1118	+	`result is 83.59% and passes the enforced 83% floor. The release smoke also`
	1119	+	`passes, and the inspected local receipt records zero automatic retries.`
1107	1120
1108	1121		`## A2. Blocker: staging is not isolated from production today`
1109	1122