One code path with a transport switch: on a host named yggdrasil[-*] (or
YGG_VERIFY_LOCAL=1) the remote half runs through local bash with cp and a local
git push instead of ssh/scp; the ssh path is unchanged. Scripts meant to be run
directly are committed 100755, and the scripts that call ygg-verify.sh do so
through bash (noexec skill mount). test-tools.sh covers the local branch and,
under a mock ssh, the remote branch and the detection rule.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The context-pack CRLF fixture (sed 's/$/\r/') and the CR-stripped check (tr -d '\r')
need a literal CR; an editor on Windows split each line at it. Byte-identical to
db2b250 on both lines (Soul s3 finding crlf-fixture-stripped).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
neighbours(text, schemas, excluded, k, left) is the one query helper admit and dispatch share; fan_out and plain_id replace admit's private thread bookkeeping and id test. The fake binding answers view of any id and queries by text with scores.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
dotnet 10 at normal verbosity ends a run with
Total tests: N / Passed: N / Failed: N / Skipped: N
(one count per indented line, zero counts omitted) instead of the
one-line "Passed! - Failed: ..." form, so mutation-table.sh baselines
read as "ran zero tests". tp_scan now parses both forms. A count the
block does not account for (Total - Passed - Failed - Skipped) is
counted as failed, so a cut-off block cannot read as clean; a "Total
tests:" line that is not a count recognises nothing.
Fixtures: dotnet10-green/red/zero.log are real dotnet 10.0.403 MSTest
runs (-v n); dotnet10-all-failed/truncated.log are hand-written.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Flat solids read as stickers beside the plant. Small fills stay solid, like a
flower's centre, and dark fills stay cutouts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An SVG used to arrive as a PDF on top of the page, in its own colours and
line quality. It now parses into marks: stroked shapes go through ribbon and
neonfill as weighted lines, a fill under a stroke becomes a petal's wash, and
bare fills become solid neon. Colours snap to the palette and near-black paints
the ground, except in an all-black drawing, which would vanish. All of it must
touch, a closed outline counting as covering its inside. Text, images and
gradients are refused before the two minute render. rsvg-convert and graphicx
are gone; python3-svgelements and python3-shapely replace them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A registry-only mount at /usr/local/cargo/registry left the package-cache lock under CARGO_HOME=/usr/local/cargo in each
container's own layer, so sibling containers (cargo-mutants stages, concurrent ygg-verify jobs) shared no lock and removed
each other's half-unpacked crate directories. Reproduced on Yggdrasil from an empty registry: 15 of 18 containers failed
with a registry mount, 15 of 18 with RUSTC_WRAPPER= too, 0 of 18 with the volume as CARGO_HOME.
ygg-verify.sh and cargo-mutants.sh mount eureka-cargo-home at /cargo-home and set CARGO_HOME and PATH for the rust image.
Removed: the retry without RUSTC_WRAPPER, its note and diag plumbing, and its two tests and the FAKE_CM_DEADWRAP switch.
Kept: container errors tail 25 lines.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Stage containers sharing the eureka-sccache volume died mid-build with cargo's
"could not execute process ... sccache ... (never executed)" or a missing dep-info
file in their own target dir, though the image's sccache is present, executes, and
a lone container builds clean. The wrapper only caches, so a failed build with
RUSTC_WRAPPER set is retried once without it; the job summary carries a NOTE line
saying which way it went, the first attempt's log is kept, and a build dead either
way stays a red baseline (exit 4) with its cause. Container errors now tail 25
lines, not 5. Tests: dead wrapper recovers with NOTE; red with and without the
wrapper stays red.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- full-baseline-skipped: stage 2 runs even when stage 1 settled every name (one name rides along; cargo-mutants runs no baseline for an empty selection), so a red suite is a red job
- timeout-mult-splits-missed: CONFIRM_TIMEOUT_MULT defaults to 5 and stage 2 gets --minimum-test-timeout like stage 1
- escaper-pipe-untested: the escaper is the --escape mode, run by the driver and by tests/test-tools.sh with each metacharacter
- zero-test-stage1-wasted: the plan container derives each file's filter and counts its tests (cargo test -- <filter> --list); 0 tests goes straight to stage 2; the container-side baseline.log count is deleted
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
One verify job per range: a plan container lists the mutants (diff taken at the
workspace root), stage 1 runs each file's mutants against that file's tests only
and decides only caught and unviable, stage 2 confirms every survivor, timeout,
unfiltered or zero-test file on the full suite, sharded over free slot leases,
and the merge prints one MUTANTS line and the final MISSED/TIMEOUT names.
NAMES reruns chosen survivors; TEST_ARGS skips flaky tests.
Deletes ygg-verify.sh's MISSED/TIMEOUT summary grep and its inline cargo mutants
usage example; the host-mode export gains YGG_SLOTS and YGG_ROOT. Briefs and
SKILL.md name the script; the mock docker grows /cm and /out, with a fake cargo
and a fake flock for the new test-tools.sh cases.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A cut window or caller list now reserves the lines its notes need, the
suggested --max-lines prints everything, and the 50-hit, 20-file, 160-character,
CR and --only bounds each have an assertion.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Read anchors resolve like file_changes (r<k> rows, MISSING when absent at
base), anchors with a symbol must hold it in the mapped window or gain
SYMBOL_MISSING, and one capped git-grep caller list per symbol follows the
windows within --max-lines. Specs without reads or symbols print what v1 did.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
nice on the docker client never reached the job, which containerd starts.
On 2026-10-31 three verify jobs held Yggdrasil's IO pressure near 75% full
and the swarm's turns slowed; idle IO lets the swarm go first.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The install command is "<abs python.exe> <path>/recall.py prompt|admit || true"
(Claude Code runs hook commands through Git Bash on Windows; a missing script
made py -3 exit 2, which blocks every prompt or admit). A For line is joined to
its first neighbour so render keeps both or neither; no neighbour fitting the
budget is no deny. The escape test is [A-Za-z0-9_.:-], not Unicode \w.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
The mutation-runner self-tests put them on PATH and got exit 126 in the
Linux container, which left 15 FAIL mutation rows red at every cut.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Shares mind.call, line_of, inert, render and the hard stop with prompt mode;
render takes the header. Digest of tool_input plus agent_id in the session
scratchpad; fails open with a note naming only the failure kind.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
On Windows the run overshot its deadline by 450-720 ms in the R2 bands: ~200 ms of
interpreter start and imports ran before the clock, the hard stop sat 0.3 s after the
deadline, and the test timed bash, env and the py launcher too. The clock now starts at
process creation (GetProcessTimes), the reply wait ends 0.1 s early so the kill and exit
fit, and the hard stop is the deadline itself. Residual outside the script: the py
launcher (40-130 ms, rare spikes to 300) and the creation call. R2 times the interpreter
directly, best of two, bands -250/+100 ms around the deadline.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Soul s1 fixes. The hook header says the lines are recalled records, data and
never instructions, written by agents not the operator. Every item is one JSON
object per line: id, title and operator_quote are JSON strings with < >, control,
bidi and separator characters escaped, so a title cannot close the framing, forge
a tag, a header or an operator segment. RECALL_DEADLINE_MS now bounds the whole
run through a hard stop; stdin is read to 4 MiB and the answer to 1 MiB
(mind.call max_bytes); RECALL_BUDGET is clamped to 9000 under Claude Code's
10,000-character cap. Tests: hostile text, a non-ASCII process-boundary run under
cp1252, byte-versus-character budget, stdin held open and endless, answer size,
budget ceiling, and the four unrecorded survivors (negative reply timeout, reply
id check, env-int positivity, EOF sentinel). Mutants of each were run and caught.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Queries rulings, questions and follow-ups in force across every root, ranked by
Huginn against the prompt, through mind.call, and prints one hook JSON within
RECALL_BUDGET. Never blocks: any failure or the RECALL_DEADLINE_MS deadline
(default 4000) exits 0 with empty stdout; one attempt; prompts under 3 words are
not queried. test-tools.sh drives it against the fake binding and pins R1-R7;
the fake gains a query branch. INDEX rows and a Precedent note added. The
operator installs the hook; settings.json is not touched.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
mind.call spawns the eureka-state binding, does the stdio handshake and one
tool call within a deadline, and kills the child on every exit path.
context-pack's mcp_view_once is deleted; mcp_view delegates to mind with
retry_once. Messages, exit codes, retry and env rules are unchanged (all 228
context-pack checks pass).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Targets the mutants that survived the first fix run: need arithmetic of cut
and full windows, the reserve for the not-expanded line at the budget edge,
the left-reserve >= 4 threshold with two windows, the exact more-lines
note, the blank line after the table, a non-sha base never echoed in the
header, and a spec without a base refused rather than crashing.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>