<?xml version="1.0" encoding="utf-8"?>
<entry xmlns="http://www.w3.org/2005/Atom" xml:base="https://docs.spec-kitty.ai/adr/3.x/2026-08-13-6-gate-outcomes-carry-severity-operator-strategy-decides-effect/">
  <id>https://docs.spec-kitty.ai/adr/3.x/2026-08-13-6-gate-outcomes-carry-severity-operator-strategy-decides-effect/</id>
  <title>ADR: Gate outcomes carry a typed severity; an operator-configured error-handling strategy decides the CLI effect</title>
  <updated>2026-11-12T00:00:00.000Z</updated>
  <author><name>Spec Kitty</name></author>
  <link rel="alternate" type="text/html" href="https://docs.spec-kitty.ai/adr/3.x/2026-08-13-6-gate-outcomes-carry-severity-operator-strategy-decides-effect/"/>
  <summary>Gate outcomes carry a typed severity; an operator strategy maps it to the CLI effect. The default blocks at BLOCKING only; trust failures are always BLOCKING.</summary>
  <content type="html">&lt;h2 id=&quot;gate-outcomes-carry-a-typed-severity-an-operator-configured-strategy-decides-the-cli-effect&quot;&gt;Gate outcomes carry a typed severity; an operator-configured strategy decides the CLI effect&lt;/h2&gt;
&lt;p&gt;&lt;strong&gt;Filename:&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;2026-08-13-6-gate-outcomes-carry-severity-operator-strategy-decides-effect.md&lt;/code&gt;&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Status:&lt;/strong&gt; Accepted (2026-09-27, amended in place; amended 2026-11-12)&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Date:&lt;/strong&gt; 2026-08-13 (proposed) · 2026-09-27 (amended and accepted) · 2026-11-12 (amended)&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Deciders:&lt;/strong&gt; Operator (ATDD)&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Technical Story:&lt;/strong&gt; Resolves the fail-closed (&lt;a href=&quot;/adr/3.x/2026-08-13-2-gates-are-declarative-asset-backed-doctrine-artefacts/&quot;&gt;ADR 2026-08-13-2&lt;/a&gt;)
vs skip-and-proceed (&lt;a href=&quot;/adr/3.x/2026-08-13-4-executable-doctrine-runs-only-from-trusted-publishers/&quot;&gt;ADR 2026-08-13-4&lt;/a&gt;)
contradiction the dialectics pass surfaced. Instead of a per-gate boolean disposition, gate
outcomes are typed by severity and the CLI effect is an operator policy. Implementation tracks
#3416 (minimal kernel ladder) and #3417 (strategy).&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id=&quot;amendment-note-2026-09-27&quot;&gt;Amendment note (2026-09-27)&lt;/h2&gt;
&lt;p&gt;Accepted together with ADR -2 and ADR -4 as one set, after the pack-trust ADR review
(architect-alphonso / paula-patterns squad, operator-approved 2026-09-27). The body is
&lt;strong&gt;rewritten in place&lt;/strong&gt;. Changes against the Proposed text:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Contradiction fixed.&lt;/strong&gt; The Proposed text defined &lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt; as “may proceed degraded” but
made the default &lt;code dir=&quot;auto&quot;&gt;block_above(RECOVERABLE)&lt;/code&gt;, which blocks it. The default now &lt;strong&gt;blocks at
&lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt; only&lt;/strong&gt;; &lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt; proceeds degraded and is recorded loudly.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Could-not-run severities are fixed by the dispatcher&lt;/strong&gt;, not declared by the gate. This answers
the 2026-08-15 fail-open reconcile question on #2535/#2599/#3417 with reading (a): machinery
faults are &lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt;, trust and tamper failures are &lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Guardrail added:&lt;/strong&gt; repository-committed configuration cannot demote trust or tamper outcomes
below &lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt;; the effective strategy is surfaced in every verdict.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Scoping (open question 4) resolved:&lt;/strong&gt; a minimal kernel ladder ships first; consolidating the
other Severity enums is follow-on debt (#3416), not a precursor.&lt;/li&gt;
&lt;li&gt;Stale reference fixed: &lt;code dir=&quot;auto&quot;&gt;status/doctor.py&lt;/code&gt; &lt;code dir=&quot;auto&quot;&gt;Severity&lt;/code&gt; is now at line 33.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id=&quot;amendment-2026-11-12-a-proven-foreign-evidence-tree-refuses-move-task&quot;&gt;Amendment (2026-11-12): a proven foreign evidence tree refuses &lt;code dir=&quot;auto&quot;&gt;move-task&lt;/code&gt;&lt;/h2&gt;
&lt;p&gt;Operator ruling 2026-11-12 (decision &lt;code dir=&quot;auto&quot;&gt;01M7733YYK1615ERKAY41C2M7C&lt;/code&gt;, Mission
&lt;code dir=&quot;auto&quot;&gt;lane-code-verification-01M76AC0&lt;/code&gt;, #7554, #7547). Before this, a test run inside a lane
worktree could load the repository root checkout’s source through its editable install, so
the pre-review gate could pass a work package on code that was not the lane’s. The gate now
runs its tests through the lane’s own test invocation and probes, before it runs, which tree
that invocation imports. This amendment records how the gate treats the probe’s answer.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;New outcome &lt;code dir=&quot;auto&quot;&gt;foreign_tree&lt;/code&gt;, a provenance refusal.&lt;/strong&gt; When the probe proves the run would
load code from another tree, the gate returns &lt;code dir=&quot;auto&quot;&gt;foreign_tree&lt;/code&gt; and runs no test.
&lt;code dir=&quot;auto&quot;&gt;move-task &amp;#x3C;WP&gt; --to for_review&lt;/code&gt; refuses with &lt;code dir=&quot;auto&quot;&gt;EVIDENCE_TREE_MISMATCH&lt;/code&gt; and exit 1, and the
transition is not applied. This holds &lt;strong&gt;whatever &lt;code dir=&quot;auto&quot;&gt;review.fail_on_pre_review_regression&lt;/code&gt;
says&lt;/strong&gt;: that setting governs new test failures, and a foreign tree is not a test failure but
evidence that no lane test result exists.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;&lt;code dir=&quot;auto&quot;&gt;--force&lt;/code&gt; bypasses it, on the record.&lt;/strong&gt; A forced move proceeds and records
&lt;code dir=&quot;auto&quot;&gt;force_bypassed: true&lt;/code&gt; and &lt;code dir=&quot;auto&quot;&gt;bypassed_refusal: &quot;EVIDENCE_TREE_MISMATCH&quot;&lt;/code&gt;; the outcome stays
&lt;code dir=&quot;auto&quot;&gt;foreign_tree&lt;/code&gt;, because no test ran. This is the difference from the trust failures in the
table below: pack trust, pin hash mismatch and tamper stay &lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt; with no bypass. A
foreign tree says where the code came from, not that executable doctrine is untrusted, so
the operator may proceed knowingly.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Carve-out from “gate evaluation never breaks &lt;code dir=&quot;auto&quot;&gt;move-task&lt;/code&gt;” (C-003).&lt;/strong&gt; Only a &lt;strong&gt;proven&lt;/strong&gt;
foreign tree refuses. A probe that faults or times out, or a project the probe cannot
inspect (for example one that is not a Python package), records &lt;code dir=&quot;auto&quot;&gt;unverified&lt;/code&gt; and the
transition proceeds as before. Every other machinery fault keeps the C-003 promise and maps
to &lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Probed only where tests run.&lt;/strong&gt; The probe runs, and the evidence tree is recorded in the
transition’s &lt;code dir=&quot;auto&quot;&gt;pre_review_gate.evidence_tree&lt;/code&gt; (lane path, head, dirty flag, probe result),
only when the gate is about to run the head’s tests: a declared test command or an override
scope. The baseline run is not probed or recorded. A gate arm that runs no test (an
undeclared gate stays &lt;code dir=&quot;auto&quot;&gt;skipped&lt;/code&gt;, an empty diff, no active binding) neither probes nor refuses.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The run environment changes for everyone.&lt;/strong&gt; The head run (declared command and the
override tier’s interpreter fallback) and the baseline run no longer inherit &lt;code dir=&quot;auto&quot;&gt;PYTHONPATH&lt;/code&gt;,
&lt;code dir=&quot;auto&quot;&gt;VIRTUAL_ENV&lt;/code&gt; or &lt;code dir=&quot;auto&quot;&gt;UV_PROJECT_ENVIRONMENT&lt;/code&gt; from the caller whenever the gate builds a lane
test invocation (every lane and write checkout; a planning-lane work package, or a build
that failed open, keeps the inherited environment). A tree with a
&lt;code dir=&quot;auto&quot;&gt;.venv&lt;/code&gt; of its own runs with that venv’s script directory first on &lt;code dir=&quot;auto&quot;&gt;PATH&lt;/code&gt;; a tree without
one gets &lt;code dir=&quot;auto&quot;&gt;PYTHONPATH=&amp;#x3C;tree&gt;/src&lt;/code&gt; when it has tracked &lt;code dir=&quot;auto&quot;&gt;src/&lt;/code&gt; packages. The head run also gets
the lane’s test database name; the baseline, which has no lane identity, does not. A project
whose tests relied on an inherited &lt;code dir=&quot;auto&quot;&gt;PYTHONPATH&lt;/code&gt; must declare it in its test command. On the
override tier’s &lt;code dir=&quot;auto&quot;&gt;uv run --project &amp;#x3C;lane&gt;&lt;/code&gt; path the run uses the environment uv resolves for
the lane rather than the probed one, so there the evidence describes the lane’s invocation,
not that exact process; the probe covers uv’s interpreter only when the lane has a &lt;code dir=&quot;auto&quot;&gt;.venv&lt;/code&gt;
of its own. When &lt;code dir=&quot;auto&quot;&gt;uv&lt;/code&gt; is not on the run’s &lt;code dir=&quot;auto&quot;&gt;PATH&lt;/code&gt; (looked up on the lane’s environment, not
the caller’s) or the lane has no &lt;code dir=&quot;auto&quot;&gt;uv.lock&lt;/code&gt;, the fallback runs the lane’s own &lt;code dir=&quot;auto&quot;&gt;.venv&lt;/code&gt;
interpreter when it has one, the interpreter the probe checked, and otherwise the CLI’s
interpreter with &lt;code dir=&quot;auto&quot;&gt;PYTHONPATH=&amp;#x3C;tree&gt;/src&lt;/code&gt;, which the probe also used.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;What the probe covers.&lt;/strong&gt; It checks the git-tracked &lt;code dir=&quot;auto&quot;&gt;src/&amp;#x3C;package&gt;/__init__.py&lt;/code&gt;
packages of the tree. A flat or namespace-package layout has nothing to check, so the
result is &lt;code dir=&quot;auto&quot;&gt;unverified&lt;/code&gt; and the gate does not refuse. It reports where packages load from;
it does not prove the code there is the lane’s reviewed code. The interpreter still starts
normally, so &lt;code dir=&quot;auto&quot;&gt;.pth&lt;/code&gt; files and &lt;code dir=&quot;auto&quot;&gt;sitecustomize&lt;/code&gt; in its environment run.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The skip switch still skips everything.&lt;/strong&gt; The pre-review gate’s skip flag and environment
variable bypass the gate entirely, the probe and this refusal included. The amendment makes
no stronger guarantee than that.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;The outcome table gains one row (below). The default strategy and the no-demotion guardrail
are unchanged: &lt;code dir=&quot;auto&quot;&gt;foreign_tree&lt;/code&gt; is not read from the configured threshold at all.&lt;/p&gt;
&lt;h2 id=&quot;context-and-problem-statement&quot;&gt;Context and Problem Statement&lt;/h2&gt;
&lt;p&gt;The gate design produced two conflicting per-gate booleans: gates flip &lt;strong&gt;fail-closed&lt;/strong&gt; (block on
failure), but untrusted packs were to be &lt;strong&gt;skipped&lt;/strong&gt; (proceed). Same un-run gate, opposite
outcomes, decided by &lt;em&gt;why&lt;/em&gt; it did not run — and the “skip” branch silently no-ops security gates
in CI. A single boolean cannot express “proceed, but degraded, and record it,” which is the real
operator need for many checks (a docs lint failing should not necessarily block a transition; a
missing security gate should).&lt;/p&gt;
&lt;p&gt;The live behaviour adds a constraint. The &lt;strong&gt;C-003 invariant — gate evaluation never breaks
&lt;code dir=&quot;auto&quot;&gt;move-task&lt;/code&gt;&lt;/strong&gt; — is implemented by &lt;code dir=&quot;auto&quot;&gt;_mt_fail_open_gate&lt;/code&gt;
(&lt;code dir=&quot;auto&quot;&gt;src/specify_cli/cli/commands/agent/tasks_move_task.py&lt;/code&gt;, FR-013), which turns any gate
exception into an unverified &lt;code dir=&quot;auto&quot;&gt;NO_COVERAGE&lt;/code&gt; warning. Any new model must keep that promise for
machinery faults without extending it to security failures.&lt;/p&gt;
&lt;p&gt;The codebase already has the ingredients, &lt;strong&gt;fragmented across five sites&lt;/strong&gt;:
&lt;code dir=&quot;auto&quot;&gt;kernel.glossary_types.Severity&lt;/code&gt; (&lt;code dir=&quot;auto&quot;&gt;src/kernel/glossary_types.py:70&lt;/code&gt;), &lt;code dir=&quot;auto&quot;&gt;audit.models.Severity&lt;/code&gt; +
its threshold &lt;code dir=&quot;auto&quot;&gt;fail_on: Severity | None&lt;/code&gt; with &lt;code dir=&quot;auto&quot;&gt;severity &amp;#x3C;= fail_on_threshold&lt;/code&gt;
(&lt;code dir=&quot;auto&quot;&gt;src/specify_cli/audit/models.py:47,53,203&lt;/code&gt;), &lt;code dir=&quot;auto&quot;&gt;status.doctor.Severity&lt;/code&gt;
(&lt;code dir=&quot;auto&quot;&gt;src/specify_cli/status/doctor.py:33&lt;/code&gt;), the charter-lint &lt;code dir=&quot;auto&quot;&gt;SEVERITY_ORDER&lt;/code&gt; blocking ladder
(&lt;code dir=&quot;auto&quot;&gt;src/specify_cli/charter_runtime/lint/findings.py&lt;/code&gt;, consumed by &lt;code dir=&quot;auto&quot;&gt;analysis_report.py&lt;/code&gt;), and
the &lt;code dir=&quot;auto&quot;&gt;strictness&lt;/code&gt; literals (&lt;code dir=&quot;auto&quot;&gt;kernel.glossary_types.Strictness&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;runtime … schema.py&lt;/code&gt;). The
&lt;code dir=&quot;auto&quot;&gt;audit&lt;/code&gt; module is already &lt;em&gt;exactly&lt;/em&gt; the “block above a threshold” pattern requested.&lt;/p&gt;
&lt;h2 id=&quot;decision-drivers&quot;&gt;Decision Drivers&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;One coherent model for “what happens when a check does not pass,” across failed / degraded /
informational / could-not-run outcomes.&lt;/li&gt;
&lt;li&gt;Operator owns strictness, not each gate author — mirrors linter/typechecker strictness config.&lt;/li&gt;
&lt;li&gt;Security and trust failures fail closed; machinery faults never silently pass and never break
&lt;code dir=&quot;auto&quot;&gt;move-task&lt;/code&gt; (C-003).&lt;/li&gt;
&lt;li&gt;Do not add a &lt;strong&gt;sixth&lt;/strong&gt; Severity enum; converge on one canonical ladder — without making that
convergence a blocker for the gate work.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id=&quot;decision-outcome&quot;&gt;Decision Outcome&lt;/h2&gt;
&lt;p&gt;&lt;strong&gt;A gate produces a typed outcome; the operator’s error-handling strategy maps it to a CLI
effect.&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Severity ladder&lt;/strong&gt; (&lt;code dir=&quot;auto&quot;&gt;ERROR_SEVERITY&lt;/code&gt;): &lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt; &gt; &lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt; &gt; &lt;code dir=&quot;auto&quot;&gt;WARN&lt;/code&gt; &gt; &lt;code dir=&quot;auto&quot;&gt;INFO&lt;/code&gt;.
&lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt; means “the system proceeds in a &lt;strong&gt;degraded&lt;/strong&gt; fashion”: the outcome is recorded
and surfaced loudly, and the transition is not stopped under the default strategy.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Every gate outcome carries a severity.&lt;/strong&gt; Who sets it depends on the outcome:&lt;/p&gt;








































&lt;table&gt;&lt;thead&gt;&lt;tr&gt;&lt;th&gt;Outcome&lt;/th&gt;&lt;th&gt;Severity&lt;/th&gt;&lt;th&gt;Set by&lt;/th&gt;&lt;/tr&gt;&lt;/thead&gt;&lt;tbody&gt;&lt;tr&gt;&lt;td&gt;Untrusted pack, hash mismatch against pin or consent, tamper detected at dispatch (ADR -4)&lt;/td&gt;&lt;td&gt;&lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt;&lt;/td&gt;&lt;td&gt;dispatcher (fixed)&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;Runner crash, timeout, missing interpreter, unconfinable host (ADR -2 baseline)&lt;/td&gt;&lt;td&gt;&lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt;&lt;/td&gt;&lt;td&gt;dispatcher (fixed)&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;Pre-review gate: the probe proves the lane’s tests would load code from another tree (&lt;code dir=&quot;auto&quot;&gt;foreign_tree&lt;/code&gt;, amendment 2026-11-12)&lt;/td&gt;&lt;td&gt;refusal &lt;code dir=&quot;auto&quot;&gt;EVIDENCE_TREE_MISMATCH&lt;/code&gt;; &lt;code dir=&quot;auto&quot;&gt;--force&lt;/code&gt; bypasses and is recorded&lt;/td&gt;&lt;td&gt;pre-review gate (fixed)&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;Pre-review gate: the probe faults, times out or cannot inspect the project (&lt;code dir=&quot;auto&quot;&gt;unverified&lt;/code&gt;)&lt;/td&gt;&lt;td&gt;recorded, never refuses&lt;/td&gt;&lt;td&gt;pre-review gate (fixed)&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;A detected violation (the check ran and failed)&lt;/td&gt;&lt;td&gt;the gate’s declared &lt;code dir=&quot;auto&quot;&gt;severity&lt;/code&gt;&lt;/td&gt;&lt;td&gt;gate&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;Pass&lt;/td&gt;&lt;td&gt;—&lt;/td&gt;&lt;td&gt;—&lt;/td&gt;&lt;/tr&gt;&lt;/tbody&gt;&lt;/table&gt;
&lt;p&gt;Fail-open therefore survives &lt;strong&gt;only for machinery faults&lt;/strong&gt;, and only as a recorded degraded
outcome, never as a silent pass. Trust and tamper failures fail closed. This keeps C-003 for
the grandfathered &lt;code dir=&quot;auto&quot;&gt;spec-kitty-pre-review&lt;/code&gt; code gate: its machinery faults map to
&lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt;, so &lt;code dir=&quot;auto&quot;&gt;move-task&lt;/code&gt; still completes, degraded and on the record.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;The operator selects an error-handling strategy&lt;/strong&gt; in &lt;code dir=&quot;auto&quot;&gt;.kittify&lt;/code&gt; config, e.g.
&lt;code dir=&quot;auto&quot;&gt;block_above(threshold)&lt;/code&gt; — outcomes at or above the threshold block the transition; below it,
the CLI proceeds (degraded) and records the finding. This is the existing &lt;code dir=&quot;auto&quot;&gt;audit&lt;/code&gt; &lt;code dir=&quot;auto&quot;&gt;fail_on&lt;/code&gt;
threshold pattern, generalised.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;A default strategy ships: block at &lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt; only.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt; stops a transition;
&lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt; proceeds degraded and is recorded loudly; &lt;code dir=&quot;auto&quot;&gt;WARN&lt;/code&gt; and &lt;code dir=&quot;auto&quot;&gt;INFO&lt;/code&gt; are recorded. Because
trust failures are fixed at &lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt;, security gates fail &lt;strong&gt;closed&lt;/strong&gt; out of the box, and CI
without operator trust seeding blocks loudly rather than silently no-ops.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Guardrail: no demotion of trust failures from the repository.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;.kittify/config.yaml&lt;/code&gt; is
committed, so a strategy read from repository configuration can make things stricter but can
never make trust / tamper outcomes non-blocking. Those outcomes block regardless of the
configured threshold.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Surface the effective strategy.&lt;/strong&gt; Every verdict records the effective strategy and where it
came from, so “why did this pass?” always has an answer.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Canonical severity, minimal first.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;ERROR_SEVERITY&lt;/code&gt; is defined once in the kernel (the
layering-correct home, importable by doctrine) as a &lt;strong&gt;minimal ladder for gates&lt;/strong&gt;, and the gate
schema references it. Migrating the other four Severity enums onto it is follow-on debt
(#3416 rescoped), not a precursor that blocks the gate work (#3417, #3418).&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id=&quot;consequences&quot;&gt;Consequences&lt;/h3&gt;
&lt;h4 id=&quot;positive&quot;&gt;Positive&lt;/h4&gt;
&lt;ul&gt;
&lt;li&gt;One knob for strictness; the fail-closed/skip contradiction dissolves — the &lt;em&gt;operator’s&lt;/em&gt;
threshold decides, uniformly, except for trust failures, which always block.&lt;/li&gt;
&lt;li&gt;The long-standing fail-open question has one answer that both keeps &lt;code dir=&quot;auto&quot;&gt;move-task&lt;/code&gt; unbreakable
for machinery faults and closes the silent-skip hole for security.&lt;/li&gt;
&lt;li&gt;&lt;code dir=&quot;auto&quot;&gt;RECOVERABLE&lt;/code&gt; expresses “proceed degraded,” which no boolean could.&lt;/li&gt;
&lt;li&gt;A minimal kernel ladder unblocks the gate work now and is the anchor the later consolidation
converges on.&lt;/li&gt;
&lt;/ul&gt;
&lt;h4 id=&quot;negative&quot;&gt;Negative&lt;/h4&gt;
&lt;ul&gt;
&lt;li&gt;Until the follow-on consolidation lands there are temporarily &lt;strong&gt;six&lt;/strong&gt; Severity definitions;
the kernel ladder must be the one every new consumer uses.&lt;/li&gt;
&lt;li&gt;A permissive operator strategy can still weaken &lt;em&gt;non-trust&lt;/em&gt; gates (a violation declared
&lt;code dir=&quot;auto&quot;&gt;BLOCKING&lt;/code&gt; can be demoted by the configured threshold); the effective-strategy record makes that
visible rather than preventing it.&lt;/li&gt;
&lt;/ul&gt;
&lt;h4 id=&quot;neutral&quot;&gt;Neutral&lt;/h4&gt;
&lt;ul&gt;
&lt;li&gt;The strategy grammar (&lt;code dir=&quot;auto&quot;&gt;block_above(threshold)&lt;/code&gt; vs a per-severity action map) is a config-schema
detail (open question).&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id=&quot;resolved-and-open-questions&quot;&gt;Resolved and open questions&lt;/h3&gt;
&lt;ol&gt;
&lt;li&gt;&lt;em&gt;Open:&lt;/em&gt; extend/rename &lt;code dir=&quot;auto&quot;&gt;kernel.glossary_types.Severity&lt;/code&gt;, or mint a new &lt;code dir=&quot;auto&quot;&gt;ERROR_SEVERITY&lt;/code&gt; next to
it and migrate later? (Either way it lives in the kernel.)&lt;/li&gt;
&lt;li&gt;&lt;em&gt;Open:&lt;/em&gt; strategy grammar: single &lt;code dir=&quot;auto&quot;&gt;block_above(threshold)&lt;/code&gt;, or a per-severity action map
(&lt;code dir=&quot;auto&quot;&gt;{BLOCKING: block, RECOVERABLE: warn, WARN: warn, INFO: log}&lt;/code&gt;)?&lt;/li&gt;
&lt;li&gt;&lt;em&gt;Resolved (2026-09-27):&lt;/em&gt; could-not-run severities are fixed by the dispatcher; a detected
violation carries the gate’s declared severity.&lt;/li&gt;
&lt;li&gt;&lt;em&gt;Resolved (2026-09-27):&lt;/em&gt; a minimal kernel ladder ships inside the gate work; full
consolidation is follow-on.&lt;/li&gt;
&lt;/ol&gt;</content>
</entry>
