Skip to content

Land eval-score (#146: the SyberEval safety gate, scorer, paired report, hidden_feedback and spend ledger) on top of the eval-core fixes of #148, with one hidden-copy integrity mechanism and one spend - #150

Merged
sykosyber merged 2 commits into
mainfrom
syberlabs/a248e101
Oct 29, 2026

Conversation

@sykosyber

@sykosyber sykosyber commented Oct 28, 2026 •

Copy link
Copy Markdown
Member

Objective: Land eval-score (#146: the SyberEval safety gate, scorer, paired report, hidden_feedback and spend ledger) on top of the eval-core fixes of #148, with one hidden-copy integrity mechanism and one spend ledger

Class: R
Agent-platform: claude@opus-5.5
Admitted-by: sykosyber
Reviewed-by: rules@2 (critic check critic.rules: high 0, medium 0, low 0)
Evidence: tested

Accepted in SyberLabs thread a248e101-021e-4a08-9b35-0fefe78d839a as candidate c1 (commit 098a4f323a3e942ef1660ea5e4089e9258591f31).
Host checks on that tree: conformance passed, critic.rules passed, scenarios passed, tests passed.

Reviewed-by: claude (independent verifier session, SyberFactory workflow)

…ity mechanism (tree_digests at four points), one strict spend ledger, gate/score/report, hidden_feedback

SyberLabs-Thread: a248e101-021e-4a08-9b35-0fefe78d839a
SyberLabs-Candidate: c1
SyberLabs-Provider: claude@opus-5.5
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Oct 28, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-10-28T23:21:08.392364Z 098a4f3 PR opened
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 098a4f323a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread syberlabs/eval/gate.py


def _evidence_files(ev: Evidence) -> list[Path]:
return sorted(p for p in ev.folder.iterdir() if p.is_file() and p.name not in DERIVED + BINARY)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Scan candidate contents for credential leaks

When a provider writes a credential-shaped string into a changed source file, that content is retained in repo.bundle, but _evidence_files excludes that file and the runner’s credential-value scan checks only JSON files. The gate can therefore report pass for a bundle containing a leaked credential; scan the retained candidate diff or tree as well. Secrets are meant to stay in the environment.

AGENTS.md reference: AGENTS.md:L210-L211

Useful? React with 👍 / 👎.

Comment thread syberlabs/eval/tasks.py
for folder, dirs, files in os.walk(root):
for name in dirs:
try:
os.chmod(os.path.join(folder, name), stat.S_IRWXU)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Avoid chmodding candidate-created symlink targets

On POSIX with the default keep_work=False, if a hidden check imports candidate code that creates a symlink inside the staged copy to an external directory owned by the same user, os.walk lists it in dirs and os.chmod follows it. Cleanup then changes permissions on that external directory even though it should only release the staged copy; skip symlinks before chmod. This is possible because checks execute candidate code with host permissions, not in a sandbox.

AGENTS.md reference: AGENTS.md:L212-L215

Useful? React with 👍 / 👎.

Comment thread syberlabs/eval/report.py
Comment on lines +193 to +196
found = read_json(folder / "score.json", "score_unreadable")
if found.get("bundle_digest") != bundle_digest:
unscored.append(item.run_id)
continue

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Validate score files against their indexed digests

If someone edits score.json after scoring but leaves its bundle_digest unchanged, this check accepts it and the report uses its altered metrics or gate result. index.json already records the score digest, but the report never compares against it, so a modified score can change the reported experiment outcome without an error.

Useful? React with 👍 / 👎.

Comment thread syberlabs/eval/score.py
if bundle.get("spec_digest") != registration["spec_digest"]:
raise Rejected("bundle_mismatch", f"{path} was written under another spec digest")
found = score_run(root, folder)
write_json(folder / "score.json", found)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Allow scored bundles when resuming an eval

After eval score writes score.json beside each bundle, EvalRun.verify still requires the directory contents to match bundle.files exactly and treats score.json as an unexpected file. Since eval run verifies existing bundles before skipping them, trying to continue or rerun after scoring stops with bundle_tampered; exclude this derived file from that verification.

Useful? React with 👍 / 👎.

@sykosyber

Copy link
Copy Markdown
Member Author

Scope: owner direction in session 2026-10-28 (SyberFactory, packet eval-score-2). It lands the eval-score work of #146 (head ae117c1f72, thread eff81f77) on top of Seth Carlson's eval-core fixes, #148 (merged as 4c65aa0b8e, a rebuild of #144). It applies the lead's reconciliation decisions, and #148's verification findings 2 to 4: #148 (comment).

Packet eval-score-2, thread a248e101, candidate c1 (head 098a4f323a, tree a808c07643).

Reconciliation decisions, as applied

  1. Hidden-copy integrity: Rebuild of #144 onto main: close eleven SyberEval core holes (author: Seth Carlson) #148's mechanism and outcome shape.
  2. Spend: one ledger.
  3. Everything else from Add the SyberEval safety gate, scorer and paired report, with the hidden_feedback spec field, hidden-check integrity re-hashing and an append-only spend ledger. Each is computed from the journal and t #146 lands as built.
  4. Small fixes from Rebuild of #144 onto main: close eleven SyberEval core holes (author: Seth Carlson) #148's verification.
    • Gap 16's test has @unittest.skipUnless(hasattr(signal, "SIGKILL")) above its expectedFailure. It is skipped on Windows and is still an expected failure on Linux.
    • spec.check refuses two githits arms whose arms.githits_arm labels collide. Test: ClaudeCodeArms.test_two_githits_arms_whose_agent_labels_collide_are_refused.
      • Cases: a.b against a-b, and two 18-character names that share their first 16 characters.
      • The same names pass when only one arm has githits.
  5. GAPS. Add a TypeSafe-backed branch policy jev_api behind the BranchPolicy seam that asks TypeSafe Jev to choose among the local JevPolicy's offered actions from ids and numbers only, validates every answer, #147's TypeSafe Jev entry stays 61, and Add the SyberEval safety gate, scorer and paired report, with the hidden_feedback spec field, hidden-check integrity re-hashing and an append-only spend ledger. Each is computed from the journal and t #146's eval-score entry is now 62.
  6. Tests: Rebuild of #144 onto main: close eleven SyberEval core holes (author: Seth Carlson) #148's gaps.
    • Gaps 12, 14, 15 and 16 stay expectedFailure.
    • Gap 13 now passes under status and none. It is split into two tests:
      • test_a_candidate_cannot_carry_a_hidden_check_into_its_next_request stays expectedFailure under tails and still documents the leak;
      • test_under_status_or_none_a_candidate_cannot_carry_a_hidden_check_into_its_next_request passes. Both modes run as subtests. Each first asserts that the candidate's print reached the check's tail in the journal, and that a repair request followed, so the pass is not vacuous. It then asserts that neither canary is in request 2.

Other docs changes (docs/EVAL.md)

  • The spec: finite numbers, run-id collisions, githits label collisions, strict JSON.
  • Hidden checks: tree_digests/hidden_changes, .pytest_cache, the hidden_integrity fields, the not_acceptable/hidden_tampered outcome.
  • The githits arm row: session and agent labels.
  • Resuming: EvalRun.verify, bundle_tampered, .partial skipped, the lock, the per-run preflight.
  • The ledger: read strictly; the report refuses an uncountable ledger.
  • The gate table: the hidden_tampered row.

Files (19):

  • new: syberlabs/eval/gate.py, syberlabs/eval/score.py, syberlabs/eval/report.py, tests/test_eval_score.py;
  • changed: syberlabs/eval/{__init__,arms,cli,runner,spec,tasks}.py, syberlabs/providers.py (hook only), tests/test_eval_adversarial.py, tests/test_eval_core.py, tests/test_eval_runner.py, docs/EVAL.md, spec/GAPS.md, pyproject.toml (packages line), evals/e5-single-vs-repair/spec.json;
  • deleted: evals/e5-single-vs-repair/registration.json.

Size: 257,543 diff bytes. There is no change to the contract schema, the event kinds, the golden traces or syberlabs/cli.py.

Commands and results

  • Targeted, local, in the c1 clone (Windows 11, CPython 3.12.6): tests.test_eval_core tests.test_eval_runner tests.test_eval_score tests.test_eval_adversarial tests.test_gaps tests.test_providers tests.test_package. Result: Ran 88 tests in 143.438s, OK (skipped=1, expected failures=4). The skip is gap 16 on Windows.

  • Neighbouring modules, local, on the same tree: test_providers test_claude_code test_package test_metrics test_experiment test_evolve test_evolution. Result: Ran 165 tests in 421.530s, OK.

  • Full suite, GitHub Actions, scratch branch syberlabs/a248e101-ci (56edcd8018, tree a808c07643, the same as c1): https://github.com/SyberLabs/SyberWork/actions/runs/40783915488

    • First attempt: test (3.11), test (3.12) and test (3.13) each Ran 1229 tests, OK (skipped=2, expected failures=5). G8 strict (3.12) OK (skipped=2, expected failures=5). compose success.
    • windows failed once, with errors=1: test_console.ConsolePreview.test_preview_appends_nothing_for_anyone, ConnectionAbortedError: [WinError 10053] in an HTTP call. That module is untouched by this packet. G8 failed because it requires every leg.
    • I re-ran the failed jobs (attempt 2). Every job passed: windows Ran 1229 tests in 954.750s, OK (skipped=26, expected failures=4), and both G8 jobs succeeded. The run is green.
  • Bite, GitHub Actions, scratch branch syberlabs/a248e101-bite (ec1f14e1ab: only the four test files, on 4c65aa0b8e): https://github.com/SyberLabs/SyberWork/actions/runs/40783915123, failure, as intended.

    • Every Linux leg: Ran 1212 tests, FAILED (failures=5, errors=31, skipped=2, expected failures=5). windows: the same counts with skipped=26, expected failures=4. compose passes.
    • test_eval_score fails to import, because gate, report and score do not exist on the base.
    • Errors, mostly because the base schema refuses hidden_feedback:
      • test_eval_core: SpecRegistration (3), Tasks.test_hidden_checks_are_staged…, Preflight (2), CommittedEvals.test_e5…;
      • test_eval_runner: FeedbackHook (2), FeedbackModes, HiddenTamper, SpendLedger (5, the new strict-ledger test included), HiddenChecks (2), EndToEnd (2);
      • test_eval_adversarial: Bundles (4), Budget (2), Preflight, HiddenChecks tamper, and the new status/none leak test (2 subtests).
    • Failures:
      • SpecRegistration.test_the_schema_and_the_cross_field_rules;
      • test_the_declared_set_digest_must_be_the_tasks_as_they_are;
      • PreRegistration.test_two_cells_that_would_share_a_run_id_are_refused;
      • ClaudeCodeArms.test_each_run_of_the_githits_arm_has_its_own_githits_session;
      • the new test_two_githits_arms_whose_agent_labels_collide_are_refused.
  • Local runs:

    • python -m conformance.run: matched 23 golden traces, exit 0.
    • python benchmarks/run_scenarios.py: scenarios: 11 replayed, exit 0, digest b41f7cbb2e70…, unchanged.
    • python -m conformance.clean_install: exit 0, 50.6 s.
  • Thread checks on c1 (19 files, 257,543 diff bytes):

    Check State Time Output
    conformance passed 21.2 s matched 23 golden traces
    critic.rules passed 1.3 s "no findings over 19 changed path(s)"
    scenarios passed 0.4 s digest b41f7cbb…
    tests passed 393.0 s Ran 1229 tests, OK (skipped=26, expected failures=4) (Windows host, 8 shards)
  • certify c1: result 098a4f323a (tree a808c07643) on repo-change.v5, class R; four checks passed; chain head 618b8aad6c at 14 events.

    • Key 4b6ae2deea: before signing, I checked that the public key of $SYBERLABS_CERTIFICATE_SEED has prefix 4b6ae2deea and matches the factory's test-integrity-2-74e7d4ac.cert.pub. The written .pub is byte-identical to it.
  • verify against the factory .pub with --repo .:

    • signature ok, ed25519 4b6ae2deea;
    • repository ok: base 4c65aa0b8e and result present, 19 changed paths, the diff digest matches, and the target is at 098a4f323a;
    • journal ok: 14 of 14 events;
    • verdict pass.
  • PR CI: all 25 checks on Land eval-score (#146: the SyberEval safety gate, scorer, paired report, hidden_feedback and spend ledger) on top of the eval-core fixes of #148, with one hidden-copy integrity mechanism and one spend #150 at 098a4f323a pass (gh pr checks 150): test 3.11/3.12/3.13, windows, compose, G8 and G8 strict (each for push and pull_request), admit, policy, G2, G3, G4 and G6 on both events, and G5.

Claim ceiling: implemented and tested, locally on Windows and on CI (Linux 3.11–3.13 and Windows), with the simulated and stand-in adapters. Not measured: no eval with a model has been run, E5 included.

Known limitations

Unresolved questions

  1. The report's refusal of an uncountable ledger: does the lead accept it, or want a report that states the spend as unknown?
  2. The scratch branches syberlabs/a248e101-ci and syberlabs/a248e101-bite are left for the orchestrator to delete.

Next permissible packet: eval-safety-fixtures (GAPS 62 (1)). After merge, the lead registers E5 on a clean checkout of main, runs eval preflight, and runs it.

Links: supersedes #146 (#146), built on #148 (#148).

@sykosyber

Copy link
Copy Markdown
Member Author

Independent verification: #150 (eval-score-2) at 098a4f323a

Verdict: pass, with findings. All six reconciliation decisions are in the code. The tests, conformance, CI and the certificate all check out, and a rebuild onto current main is clean. None of the findings blocks the merge. Findings 1 and 2 are small fixes to the spend ledger: the code does not yet do what EVAL.md and the handoff claim. The lead decides whether to fix them before merge or in a follow-up.

Reviewed: head 098a4f323a3e942ef1660ea5e4089e9258591f31, tree a808c07643, base 4c65aa0b8e. I worked in a fresh clone on Windows 11 with CPython 3.12.6. I wrote no new attack code, made no model, Jev or GitHits calls, and pushed nothing.

Findings, most important first

  1. Medium: the report adds up a ledger line that names an unpriced model, and EVAL.md says it refuses one. (syberlabs/eval/report.py:234-241, against docs/EVAL.md:148)
    • report.build refuses only lines that cannot be read or whose dollars are not finite.
    • A line with finite dollars but a non-empty unknown_models is added in as if it were complete. The runner stops eval_spend_unknown on that same line (runner.py, EvalRun.spent).
    • EVAL.md:148 lists three kinds of bad line: not a record, non-finite dollars, and an unpriced model. It says they stop the eval and that "eval report refuses such a ledger with the same code".
    • Result: a --partial report of an eval that stopped on an unpriced model gives a ledger_dollars that leaves out the unpriced spend, and nothing flags it.
    • Fix: add line.get("unknown_models") to the refusal condition at report.py:235-236, and add one case to EndToEnd in test_eval_score.
  2. Low to medium: a very large JSON integer raises a bare OverflowError, not eval_spend_unknown. (syberlabs/eval/runner.py:254)
    • ledger_dollars checks type(value) in (int, float) and then calls math.isfinite(value).
    • spec.loads accepts a 401-digit integer: it is under Python's 4,300-digit limit, and it is an int, not inf. math.isfinite on that integer raises OverflowError: int too large to convert to float. I checked this with spec.loads and math.isfinite alone; I did not run an attack.
    • The error reaches EvalRun.spent (runner.py:473 and 506) and report.build (report.py:236) as an uncoded exception. It still fails closed, because nothing runs.
    • This contradicts decision 2 and the handoff: "anything else … stops the eval with the coded eval_spend_unknown, never a bare ValueError". The 12-line test covers 1e999, which is a float, but not a large integer.
    • Fix: wrap the conversion in try/except OverflowError and return None, and add one subtest line.
  3. Low: "after_check (every check phase)" is met per phase, not per evaluation. (runner.py:567-576, build.py:1177, docs/EVAL.md:86)
    • The repair-feedback evaluations run inside thread.propose. One hash covers them all together: after_propose. No hash follows each repair evaluation.
    • Thread.check re-runs checks only when the chosen candidate's evaluation is none or stale (build.py:1177). A repair or single arm's chosen candidate was already evaluated, and is fresh, during the proposal. So after_check usually follows a check that ran nothing.
    • The limit itself is stated honestly: EVAL.md:100 says "the evaluations inside one proposal are covered together", and GAPS 62 (5) says the same.
    • But EVAL.md:86 describes the third point as "after the host's check of the chosen candidate", which reads as though candidate code ran just before it.
    • This is not a regression: Rebuild of #144 onto main: close eleven SyberEval core holes (author: Seth Carlson) #148 hashed once, after the check.
    • If the lead wants a hash after each evaluation, feedback_filter is already called once per evaluated candidate and is a natural place for it. Otherwise, a one-line change to EVAL.md:86 would do.
  4. Low: the hidden_tampered rule passes on whatever hashes are present. (syberlabs/eval/gate.py:467-482, score.py:148)
    • The rule does not require after_propose, nor before_accept when the outcome is accepted. The runner always writes both, so this matters only for a bundle edited by hand.
    • score.py:148 reads fewer signals than the gate: it ignores the per-hash match and a staged digest that differs from expected. The handoff's "score marks quality untrusted on the same signals" is therefore close but not exact.
  5. Info: the min(per-run, remaining) ceiling never comes into play. (runner.py:503)
    • spent is recomputed just before the guard, and the guard requires spent + per-run <= total. So the ceiling always equals the per-run dollars, and the lock rules out "the ledger grew from elsewhere" within one runner.
    • The code is harmless. EVAL.md:148's "lower … only when the ledger grew from elsewhere" describes a case that cannot happen now.

Each decision

# Decision Result
1 Hidden integrity Applied. tasks.tree_digests and hidden_changes are the only integrity code. Integrity, staged_files and copy_digest are gone (grep finds nothing). runner.hidden_hash runs at staged, after_propose, after_check and before_accept. A change gives not_acceptable / hidden_tampered, hidden_intact: false and hidden_changed. Every hash is in bundle.hidden_integrity. The kit home and the hidden copy each get a random 128-bit name under --hidden-root, outside the work directory, and the copy is staged read-only (tasks._read_only). The before-accept hash: thread.accept is called in one place only (runner.py:576), straight after intact("before_accept"), and Thread.accept runs no checks (build.py:1228). The repair-feedback checks are covered by after_propose, not after_check (finding 3).
2 Spend Applied, with findings 1 and 2. There is one ledger, evals/<id>/spend.jsonl. runs/failed.jsonl, FAILED, EvalRun.failed and _failed are removed. _charge writes the line, flushes and fsyncs it before anything is deleted, on both the completed and the aborted path (it runs before the .partial rmtree and before the finally cleanup of the work directory, kit home and copy). The ceiling is computed as min(per-run, total - spent). ledger() uses spec.loads, and ledger_dollars rejects bool, NaN, inf and negative values. #148's test_a_run_that_fails_after_it_spent_is_charged_against_the_cap is unchanged and passes.
3 The rest of #146 Applied. hidden_feedback is a required spec field and part of the spec's digest. feedback_filter defaults to None, and the request stays byte-identical (FeedbackHook.test_without_a_filter…). An evolve arm is refused under any mode but tails. Every gate.py rule returns pass, fail, not_computable with a reason, or not_applicable; evaluate gives incomplete on any not_computable, never pass. score.py and report.py use Fraction, an exact sign test and a seeded bootstrap, and refuse runs_missing without --partial (report.py:179). The CLI has score, report --partial and --hidden-root. pyproject.toml adds syberlabs.eval. E5 has hidden_feedback: tails, and its registration.json is deleted. #148's lock, EvalRun.verify, per-run preflight, thread_label, finite numbers, run-id collision check and strict read_json are all still there, and their tests pass.
4 Small fixes Applied. Gap 16 has @skipUnless(hasattr(signal, "SIGKILL")) above @expectedFailure (test_eval_adversarial.py:267). spec.check refuses two githits arms with colliding labels. The test covers a.b/a-b and the 16-character prefix case, and also shows that the same names pass when only one arm has githits.
5 GAPS Applied. 61 is still #147's TypeSafe Jev entry, and eval-score is now 62. EVAL.md refers to 62 at lines 3 and 216. Entry 62 drops the __pycache__ claim, and EVAL.md:102 cites PYTHONDONTWRITEBYTECODE=1.
6 Gap 13 Applied. It is split in two. tails stays expectedFailure. The new status/none test runs both modes as subtests. Each first asserts that both canaries are in the journal's check tail and that a second request followed, so it cannot pass vacuously. It then asserts that neither canary is in request 2. The test passes.

Scope: 19 files, 2,650 lines added and 205 removed, 257,543 diff bytes. No contract schema, event kind, golden trace, conformance/ or syberlabs/cli.py change. The only spec/ file touched is spec/GAPS.md. Since 4c65aa0b8e, main has gained only #149 (docs/EVOLUTION.md, syberlabs/jev_api.py, tests/test_jev_api.py).

Open question 1 (should eval report refuse an uncountable ledger?): yes, that is a sound default.

  • The report states the eval's spend and whether the cap held. Leaving a line out of the sum would understate both, and a published instrument should fail closed, as eval run already does on the same line.
  • Two refinements:
    • extend the refusal to unknown_models lines (finding 1), so the report and the runner agree on what "uncountable" means;
    • later, if needed, a --partial report could print the spend as "unknown (lines …)" instead of refusing, but only under the partial label.

Tests and CI

  • Local, fresh clone at 098a4f323a: python -m unittest tests.test_eval_core tests.test_eval_runner tests.test_eval_score tests.test_eval_adversarial tests.test_providers tests.test_gaps tests.test_package gave Ran 88 tests in 138.080s, OK (skipped=1, expected failures=4). The skip is gap 16 on Windows.
  • Local: python -m conformance.run gave matched 23 golden traces, exit 0.
  • Full CI run 40783915488, on scratch branch 56edcd8018, whose tree a808c07643 is identical to the head's:
    • attempt 2: every job succeeded;
    • attempt 1: windows failed with errors=1 (Ran 1229 tests, FAILED (errors=1, skipped=26, expected failures=4)). The failure is test_console.ConsolePreview.test_preview_appends_nothing_for_anyone, a ConnectionAbortedError: [WinError 10053] in urllib against the loopback console server. The packet does not touch test_console, which does not use the eval package. G8 failed only because it needs every leg. I agree it is unrelated.
  • Bite run 40783915123, on ec1f14e1ab: four test files on top of 4c65aa0b8e, and those files are identical to the head's (git diff --quiet). Result: failure, as intended. Every Linux leg and G8 strict: Ran 1212 tests, FAILED (failures=5, errors=31, skipped=2, expected failures=5). windows: the same, with skipped=26, expected failures=4. compose passes.
  • gh pr checks 150: all 25 checks pass.

Certificate

syberlabs verify --public test-integrity-2-74e7d4ac.cert.pub --repo . eval-score-2-a248e101.cert.json, run in the fresh clone:

  • signature ok ed25519 4b6ae2deea, thread a248e101, candidate c1, result 098a4f323a (repo-change.v5, class R);
  • repository ok: base and result present, trees match, 19 changed paths, and the diff digest matches;
  • journal skipped, because a fresh clone has no .syberlabs journal;
  • verdict pass (journal not checked).

The receipt's .cert.pub is byte-identical to the factory key.

Rebuild preview

I cherry-picked 098a4f323a onto origin/main at 5ebbbce6e0, locally only.

Nothing was pushed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@sykosyber
sykosyber merged commit b2763fb into main Oct 29, 2026
25 checks passed
sykosyber added a commit that referenced this pull request Oct 29, 2026
… report refuses an unpriced ledger line, (#151)

Objective: Close the independent verifier's findings on #150 (eval-score-3): the report refuses an unpriced ledger line, a huge integer in the ledger is eval_spend_unknown, the hidden_tampered rule requires the runner's hash points, and EVAL.md states the after_check coverage and the spend ceiling exactly

Class: R
Agent-platform: claude@opus-5.5
Admitted-by: sykosyber
Reviewed-by: rules@2 (critic check critic.rules: high 0, medium 0, low 0)
Evidence: tested

Accepted in SyberLabs thread `c73912bf-1c64-4343-a513-71db9b15311d` as candidate `c1` (commit `08252d39be00b459dfef6b08fffaac92c0b7ad53`).
Host checks on that tree: conformance passed, critic.rules passed, scenarios passed, tests passed.

Reviewed-by: claude (independent verifier session, SyberFactory workflow)


Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant