test/changelog.test.sh drives changelog_fragments (order, marker, absent
dir), changelog_fragment_problem (every rule, file named each time), and
changelog_assemble (both shapes, canonical order, mixed-shape refusals).
test/changelog-assemble.test.sh drives the CLI against constructed trees:
exact-byte writes, provably read-only --check, every refusal from the
spec, the publisher/assembler round trip, and idempotence.
Closes#114.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
lib/changelog.sh gains changelog_fragments (publication order: trailing
issue number descending, filename tie-break), changelog_fragment_problem
(the release-time rules moved onto the PR that writes the fragment, #112
D9), and changelog_assemble (canonical group order per D5, one shape per
repo per D4). bin/changelog-assemble folds changelog.d/ into one release
section, deletes exactly what it consumed, and --check proves the body
without touching the tree.
Part of #112. Closes#114 groundwork; tests follow.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The unit block now feeds blind_sweep_warning a sampled reason and asserts
the new lead plus two must-fail guards: the disproven 'grant checks: read
and statuses: read' diagnosis stated as fact goes red, and so does any
drift in the counted line's whole-line shape (exactly the blind PRs match,
no more, no less — a reason line that matched would double-count, a folded
reason would undercount). read_failure_reason is covered pure: D4 wording
for empty stderr, multi-line collapse to one line, 400 chars truncated to
300 plus ellipsis within the 304-byte bound, 300 passing through whole.
blind_main_probe's gh pr view stub now fails with a denial on stderr, the
way real gh fails.
Part of #101.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The reconciler's mergeability/checks read kept its correct degrade but
threw the reason away: 2>/dev/null dropped gh's stderr, leaving a
permanent denial and a network hiccup byte-identical in the log (#95 had
to infer a cause from a control case, and the inference did not survive
incubator#48/#49). Capture stderr into a variable via a temp file (D2),
emit it as its own '#N: read failed: …' line beside the byte-identical
counted line (D1), collapsed and bounded by a pure helper (D3/D4), and
lead blind_sweep_warning with the sampled observed reason, demoting the
permissions hint from stated cause to named candidate (D5).
Part of #101 groundwork; tests and changelog follow.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The slot rule counted work in flight but had no shape for its most common
wait: the round passed, state:needs-human set, the human's merge pending.
Shape 4 names it, the handoff round summary is its declaration, and shape
2 now covers the round awaiting its first verdicts so the live and passed
rounds are sequential and non-overlapping (#109).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
CI's shellcheck gate treats info findings as red. The gh() stubs paired
with bootstrap_labels are reached only through run's "$@", which
shellcheck cannot trace (the older probes stub reconcile_pr, which calls
gh directly), so they carry reasoned SC2317 directives; the LABELS.md
doctrine parse carries SC2016 for its literal backticks. Probes now live
in named functions, matching the house *_probe() pattern.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Registry-vs-LABELS.md identity, happy path, missing label, one-name 403,
DRY_RUN narration, and the executed-subprocess dispatch (#91's lesson: a
sourced probe cannot see set -e). Cron and pull_request_target delete
nothing. Verified red without the guard: the unguarded dispatch dies on
the FIRST absent label, one delete of six attempted.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
LABELS.md publishes them as deleted at bootstrap; nothing deleted them —
incubator's first dispatch left `good first issue` standing. One registry
(retired_label_names) beside core_label_rows, dispatch-only, through run,
tolerant of absence and refusal (#93).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
reconcile_opened_issue's two early exits were bare returns, which carry
the failed guard's status into the executed script's set -e — every
triage-authored mint killed the labels run before one issue was
reconciled (#91, 4/4 observed). The stand-downs now say return 0; a
genuine failure on the arrival path still aborts loudly.
The suite sources the script and takes the set -u-only branch, so it
was blind to this by construction. The new arrival section executes the
script as a subprocess behind a fixture-serving gh stub (the house
pattern from test/release-chain.test.sh) and covers all three arrival
outcomes plus the preserved loud-failure path; it fails against
bb37c15 with the production signature — exit 1, empty output.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The shared, role-independent wake — an open issue assigned to me
carrying attention — stated once above the per-role lists, first in
priority, one acked session per demand, with the rejected mention-poll
and the #16 incident recorded and the on-paper caveat kept (#86).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The one-issue-at-a-time bullet counted claims when it meant build work
in flight; the 2026-07-23 board (#15, #16 parked beside #73) proved the
count wrong. BUILDER.md now defines the three parked shapes, the
declared-never-inferred park comment, and unparking; TRIAGE.md names a
directed hold as a park (#77).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
FLEET.md's wake conditions gain #50 D16: notify.sh's second query (open
issues and PRs labelled needs-ruling), one tracked message per item edited
in place across the ladder's four rungs and removed on clearance, the
message's contents, triage's past-24h pickup, and the on-paper caveat.
Nothing box-side sets, clears or decides the flag.
Part of #50. Closes#74.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Pure decisions (shape presence, rung boundaries, Default: parse for wording
only), sweep probes for every AC path (malformed-once, conforming silence,
rungs despite activity, cron progression, missed-moment skip, re-flag
episode, unreadable comment list, malformed+rung same pass), and the
existing nudge fixtures updated to conforming escalations with pre-seeded
rung markers so each probe observes one behavior alone.
Part of #73.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
lib/ruling.sh gains the mechanical half of #50 D12-D15: the comment body
rides the fetch as base64 (rows stay line-oriented), ruling_shape_decision
checks the four line-anchored bold-tolerant field labels,
ruling_deadline_decision derives the rung from the labeled epoch,
ruling_default_decision parses the Default: line for wording only, and
reconcile_ruling wires them with bare-stops-here exclusion — every write a
comment, one per marker per episode, both surfaces via the shared lib.
Part of #73.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
