From bbdd58da0cdac05302cf67dae2dfa63d2bdd27db Mon Sep 17 00:00:00 2001 From: tn-pisama Date: Thu, 10 Sep 2026 00:21:59 -0700 Subject: [PATCH 1/2] release: prepare pisama 0.7.0 coverage reporting migration --- CHANGELOG.md | 19 +++++++++++++++++++ pyproject.toml | 2 +- src/pisama/__init__.py | 2 +- 3 files changed, 21 insertions(+), 2 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index ac9cb64..7dee2e2 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -4,6 +4,25 @@ All notable changes to the `pisama` meta-package are documented here. The packag ## [Unreleased] +## [0.7.0] - 2026-09-10 + +### Changed + +- `check --json` uses schema 3: per-file detector assessments and coverage + completeness are included. `files_clean` requires explicit assessed coverage; + use `files_no_findings` for the old absence-of-findings count. `passed` means + the severity gate passed with no detector errors, not complete task validation. +- Explicit detector errors cause nonzero exits in analyze, check (including + `--fail-on never`), replay, smoke and watch. Unknown coverage/abstention alone + does not fail the severity gate. Consumers must update schema-2 parsers. +- `AnalyzeResult.detector_assessments` is additive. Legacy core 1.10.3 outcomes + without assessment metadata are unknown, never inferred checked passes. +- Replay comparisons retain assessments and no longer infer a fix from a + disappearing finding without comparable contract/input provenance. +- Terminal summaries distinguish findings from coverage. This release does not + itself ship the pending core communication correction or establish calibrated + confidence, representative accuracy, or successful customer outcomes. + ## [0.6.5] - 2026-09-10 ### Fixed diff --git a/pyproject.toml b/pyproject.toml index 8ecf2a3..7f800f1 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -4,7 +4,7 @@ build-backend = "hatchling.build" [project] name = "pisama" -version = "0.6.5" +version = "0.7.0" description = "Multi-agent failure detection for production AI systems" readme = "README.md" license = "MIT" diff --git a/src/pisama/__init__.py b/src/pisama/__init__.py index 45f4659..e600a30 100644 --- a/src/pisama/__init__.py +++ b/src/pisama/__init__.py @@ -14,7 +14,7 @@ __version__ = _pkg_version("pisama") except PackageNotFoundError: # running from a source checkout without an install - __version__ = "0.6.5" + __version__ = "0.7.0" from pisama._analyze import AnalyzeResult, Issue, analyze, async_analyze from pisama._http import PisamaAuthError From 7e2ec682b72332d09a188eaf21ce58e4ad54eb46 Mon Sep 17 00:00:00 2001 From: tn-pisama Date: Thu, 10 Sep 2026 00:43:03 -0700 Subject: [PATCH 2/2] docs: clarify unassessed coverage and accounting validation for 0.7.0 --- CHANGELOG.md | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 7dee2e2..ab63f89 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -8,9 +8,10 @@ All notable changes to the `pisama` meta-package are documented here. The packag ### Changed -- `check --json` uses schema 3: per-file detector assessments and coverage - completeness are included. `files_clean` requires explicit assessed coverage; - use `files_no_findings` for the old absence-of-findings count. `passed` means +- `check --json` uses schema 3: per-file assessments and + `assessment_reporting_complete` describe identified detector reports, not + whole-trace coverage. `trace_coverage` stays unassessed and `files_clean` stays + zero. Use `files_no_findings` for absence-of-findings counts. `passed` means the severity gate passed with no detector errors, not complete task validation. - Explicit detector errors cause nonzero exits in analyze, check (including `--fail-on never`), replay, smoke and watch. Unknown coverage/abstention alone @@ -22,6 +23,10 @@ All notable changes to the `pisama` meta-package are documented here. The packag - Terminal summaries distinguish findings from coverage. This release does not itself ship the pending core communication correction or establish calibrated confidence, representative accuracy, or successful customer outcomes. +- Response-contract accounting, when supplied by core, is validated against + actual trace size, record counts and detector outcome. Contradictions are + rejected. Positional checked/unsupported/outside-scope records omit raw IDs + and arbitrary metadata; they never certify whole-task business semantics. ## [0.6.5] - 2026-09-10