process_audit_verdict now calls validate_signer_separation
before any state save or ledger write (previously the function
had no production caller on this path: it only ran inside
collect_failed_checks via decide_pushback, which itself has no
production caller). Signer is bound to the authenticated
caller -- a body signer differing from the caller is rejected
-- and executor/coordinator_approver/qa_tester must be
declared and distinct from the caller; missing field,
self-audit, or forged signer raises ValueError, mapped by
_loop_run to HTTP 400 with no side effects. No new route.
Honest limits are in the helper docstring: declared-identity
enforcement only (task rows store no authoritative executor /
coordinator / QA identity; task executor is a backend, not a
person), and audit-fix reaudit verdicts are out of scope
(the fix poster is not necessarily the re-auditor, so binding
signer to the fix poster would be a mis-binding).
(cherry picked from commit 56dce5165f4d08f38fe1730b9eea3b78f90cb97d)
Reconcile-port: handler_lifecycle.py docstring hunk redirected into cloud/server.py audit-verdict handler (target keeps 6-mixin split; enforcement itself is in audit_loop_wiring.py and applied cleanly).
Pre-fix any first-attempt audit failure raised NameError at
auth.py:424 (sqlite3 never imported), aborting the 5-attempt
retry loop on attempt 1 and masking the original error. The
424-line coverage mystery is also explained: the only full-suite
hit is test_child_model.test_coordinator_inherits_child_defaults
via worker best-effort _audit swallowing an OperationalError
(database is locked); 422-423/426-427 stay uncovered because
rollback succeeds and NameError aborts before either raise.
Bubble audit (see report): no merge/release gate swallows the
NameError and proceeds, so this stays a downgrade, not a
production blocker.
(cherry picked from commit 039c6adbac3b120b5b9a1dcbafeff6a91407ce70)
Postgres audit_ledger append-only guard gains exactly one
execution entry: cloud/ledger_pg_guard.apply_pg_guard, runnable
as 'python3 -m cloud.ledger_pg_guard' (explicit ops command,
not a migration script). It executes PG_GUARD_SQL verbatim from
the constant -- no second copy of the SQL anywhere -- refuses
non-PgConnection targets, does not commit itself (caller/CLI
commits), and refuses with exit 2 on a non-postgres backend so
a no-PG environment can never claim hardening. ensure_schema
still skips PG triggers; POSTGRES_SCHEMA still builds only the
table. Tests lock all of this with a recording fake connection
(no PG execution is faked), a source scan asserting only
ledger.py and ledger_pg_guard.py reference the constant, and
CLI dry-run/refusal cases. Real-PG syntax and UPDATE/DELETE
blocking verification is explicitly left to B5, not claimed here.
(cherry picked from commit b1f6c9158d628186965a9e57576ff8c9500e3f73)
Server POST /tasks/{id}/merge no longer has record/enforce/off:
hard_mode_from_env and HARD_MODES deleted (not re-defaulted),
handler always evaluates the hard gate, env CLOUD_MERGE_GATE_MODE
and body merge_gate_mode are ignored server-side. The record
'no-evidence passes' and 'unprotected+confirm passes with
merge_gate_risk' server paths are gone with the dead risk audit;
no production caller justified keeping them (only production
caller of the non-hard evaluator was the merge handler itself;
audit_loop_wiring already used the hard gate). cloud/merge_gate
pure functions keep their #209 semantics but are documented as
non-server: they are no longer wired to any endpoint.
Server-level record/enforce/off tests rewritten to the new
behavior (renamed, not deleted silently); pure-function tests
unchanged.
(cherry picked from commit 08e75edbd6d04701965c5db951fa054cfbb75bb0)
Reconcile-port: handler_lifecycle.py hunks redirected into cloud/server.py _merge_task (target keeps 6-mixin split; mixin file does not exist here); handler_support.py kept at HEAD (its A1 hunk is a docstring of _validate_hard_gate_body, ported to server.py copy); test mock paths cloud.handler_lifecycle.* rewritten to cloud.server.*.
- restore _audit_row in cloud/automations.py as a thin wrapper over
cloud.auth.audit_on_conn (legacy row + ledger entry pair,
user_id system, best-effort self-commit preserved). The stack's
gap-4 option A deleted the direct writer; main #275 resident.py
imports _audit_row, so the bare rebase broke server import.
No direct audit_log INSERT is reintroduced, so the #250
no-bypass scan stays green.
- conformance accounting 238/144 -> 243/149: main #275 added 5
resident methods that legitimately stay in server.py; moved
count stays 94, no duplicates.
- directional: resident x3, audit chain, conformance, automations
x3 suites, 51 tests OK
- whitelist expansion (tr + exit builtin) confirmed, comment only
- B4 viewport wording: temp banner removed, wording untouched,
change-list table stays in draft, not merged into body
- B3-tail (a)/(b): no content change (bullet still absent), recorded
as decided; v2 record showed it unselected, no side picked here
- section 4: 6 templates/artifacts/ refs deleted per signed verdict;
tripwire test updated 6 -> 0 to pin the new inventory
- CHANGELOG pointer line kept, no marker present
- workers audit landing migration confirmed, comment only
- verify-script gate evidence confirmed, gate hard untouched
- 2 temp stances NOT touched: actual-only quota, evidence binding
not enforced / no anti-restamp
- behavior code unchanged; directional skill tests 39/39 OK
- cloud/evidence_binding.py: bind/validate/nonce single-use store;
release endpoint consumes nonce on first validated submission,
replay -> 409 evidence_replayed, stale/future -> 409
evidence_stale, partial -> 400, absent -> old path + label.
Binding blocks exact replay and stale evidence, NOT restamping
(unsigned evidence cannot prove content freshness).
- cloud/quota.py docstring: actual-only is the v2 integration
position (already converged in v1 base via fef09ae, NOT in
main); original PR semantics (#86 all-rows vs #90 actual-only)
still pending coordinator confirmation at merge time.
- cloud/ledger_onboarding.py: DDL checklist referencing
PG_DDL/PG_MARKER_DDL constants, read-only verify_ledger_ready
and ledger_setup_report; default stays OFF, PG DDL stays an
operator action (never claimed as executed).
- cloud/ledger_outbox.py + worker hook: ledger write failures
go to a JSONL outbox in addition to the log and can be
replayed; task outcome still unaffected (rule 7 kept),
audit fail-closed untouched -- asymmetry documented, not
silently changed.
- cloud/gate_verification.py: read-only verifier over an
injected reader (match/mismatch/skipped); live wiring in
merge/release is NOT possible on the current contract (no
repo/ref in body or task row, no guaranteed trusted token),
proven in the module header; no fake verification claimed.
- cloud/tests/test_v2_evidence.py: 18 new tests, all passing.
(cherry picked from commit bd85e33cd5e846fa6ee5d7e2440182dbd9c758b0)
- db: translate INSERT OR IGNORE ... SELECT to ON CONFLICT DO NOTHING;
register task_audit_loop_state (task_id) in _OR_REPLACE_PK (direct port).
- server: _backfill_default_quotas uses SAVEPOINT containment, no longer
commits the caller's transaction (direct port; callers are init_db only).
- worker: NOT ported — integration _coord_child_rows has no column-fallback
retry loop, so the InFailedSqlTransaction cascade site does not exist here.
- tests: rewritten for the integration base (no cloud.audit_loop /
_coord_audit_loop_merge here); real-PG cases skip without DSN/driver.
(cherry picked from commit 79e6843610fcd275af874e7b4afc57af7ffa579e)
9 failures from v1 baseline 02835f6 cleared with provisional resolutions only;
each option pair is annotated provisional in verification/code-to10-v2-failures-summary.md
and awaits user confirmation. No check deleted, no skip added, no bare-phrase
stuffing, default hard merge gate untouched. Known remainder: one pre-existing
date-dependent test_arch_converge failure (Oct-dated fixture vs Nov current
month), failing on pristine v1 too, left for the quota track.
Preparation only: no v0.3/v0.4/v0.5 text is changed, nothing is
claimed transplanted or deployed. Stacked on feat-audit-pushback
(#244, 39966bd).
- Appendix provenance 00..09: templates/roles/00-08 sources
(byte-identical to #217) plus 09-auditor extracted from #244's
appendix; check_appendix.py extended 00..08 -> 00..09.
SKILL.md sole change (1 line, format-only): 09 chapter heading
normalised to the strict source form, otherwise the #217
checker cannot recognise the chapter. check_appendix.py passes
on the real pack: 10 chapters match; --emit outputs 10 chapters.
- TRANSPLANT-PREP-244.md: per-paragraph transplant diffs for
#244's +459 lines into v0.3/v0.4/v0.5 with anchors, grades
(merge / merge+note / partial stop / full stop), chapter-number
mapping (v0.5: 11-auditor, colliding 09-ops noted), the
templates/artifacts delete-references decision with 6 sites and
replacements, and the draft-sync checklist (sync itself is
post-deployment only, not done here).
- tools/check_transplant.py: anchors / draft-apply /
check-transplanted modes; draft-apply on the real v0.3/v0.4/v0.5
texts passes check-transplanted (drafts in /tmp only, not
committed). Stopped paragraphs are listed, never silently
rewritten: v0.3 G7/full-auto sentences, v0.4 G7 sentences and
the tighten-3 pre-rebuild confirmation, v0.5 whole QA section
plus its checklist item and the confirmation paragraph.
- Tests: test_skill_appendix_provenance.py (00..09) and
test_skill_transplant_prep.py; 13 new tests pass.
Suites on this tree: cloud 1523 OK (skipped=4), top-level 105
OK, deploy 39 OK.
- cloud/attempt_hook.py (new): canonical policy -- one counted
dispatch = one row, is_defect always False from the worker;
human/first rounds and human breakpoint resumes count;
platform auto-resume reruns, coordinator fan-out, and
pre-execution gate failures do not; transient retries inside
one dispatch are not new attempts. Platform origin survives
via the task_attempt_origin_markers table (same-transaction
write on requeue, queued_at generation guard on consume).
Defects are explicit-recorder only; corrections are
append-only (record_correction, last-wins).
- cloud/worker.py: Worker(attempt_ledger_hook=False) + CLI
--attempt-ledger-hook (default OFF = gap-3 behaviour exactly);
marker write in auto_resume_interrupted, marker consume +
classify in run_one, one ledger write at each terminal point
(cancel branch, normal terminal block); all hook writes
best-effort, never fail the task or block a requeue.
- cloud/attempt_ledger.py: corrects_attempt_no column (new DDL
supersedes gap 3's pre-merge DDL) + record_correction +
correction-aware aggregation; read path uses SELECT * so
gap-3-era tables still read; cost code untouched.
- cloud/tests/test_attempt_hook.py: 32 tests (policy, markers,
corrections, worker integration, mixed endpoint).
- docs/arch-conformance-metrics.md: policy, switch, PG manual
steps (new PG_DDL + PG_MARKER_DDL), remaining user decisions.
- host.py: 7 one-shot reads switched to with open (run_health_check
pidfile, daemon-status pid/log, daemon-stop pid, daemon old pid,
interrupt poll state, parallel jobs); read content, encoding args
and error semantics unchanged.
- tests/test_host_thread_cli.py: test_old_archive_unchanged state
read switched to with open; added a warnings-capture assertion
scoped to the state read so a regression fails the test.
- Intentionally untouched: daemonize log handle (daemon-lifetime
dup2 target) and the two _load_config_file / init sites owned
by PR #266.
Cycle tasks never reach topo sort: ready_tasks filters by
'deps all completed' first, so they neither raise ValueError nor
zero out other ready tasks (5 ready + 1 cycle pair still counts 5;
only a pure cycle counts 0, because its ready set is empty).
Counting logic unchanged.
- cloud/audit_scorecard.py: pure stdlib scorer, structured evidence in,
7 dimension scores (0-10, 1 decimal), basis, missing/untested items,
overall pass only when all 7 >= 9; unknown -> None, never 0/10/estimated;
fixture/viewport/code-review evidence -> untested, never pass;
thresholds all test_default_unapproved, user_approved only on explicit input.
- server.py: GET /audit-scorecard read-only, same auth/member scope as
/metrics, no DB writes, no ledger writes, no pushback/reaudit triggers;
on main base it honestly reports unknown/not-enforced, overall not_pass.
- docs/audit-scorecard.md: spec, formulas, threshold table, unknown
handling, overall rule, 10-to-7 merge mapping, relations to
#238 metrics-report, #244 sampling, #260/#261 loop, and a system-computed
example (current state, overall not_pass) clearly separated from the
previous judgement estimates.
- tests: 12 new items pin all-pass, one-below-nine fails, average 9.1
with one 4 still fails, unknown None, fixture untested, conditional
not a pass, thresholds unapproved, endpoint 401/200 read-only.
- suites: cloud 1442 OK (skipped 4), top-level 105 OK, deploy 39 OK.
Server-side execution for the track-1 pure driver (#260), no driver
semantics changed: audit verdict / fix completion / correction /
rebuild confirmation endpoints persist LoopState in audit_loop_state,
write every step through the #241/#247 ledger path (ledger first,
state second, fail-closed), derive fix/rebuild subtasks, and block
merge/release via the existing #247 chain reader. POST
/tasks/{id}/release follows the #259 contract (global-admin write,
fresh hard-gate evidence, merged status, multi-source 409 with no
audit write on block); deploy.py stays outside task gates.
LoopState + decide_audit + after_fix + apply_correction +
confirm_rebuild + emit, all pure: the driver returns Action
objects, the caller executes (subtask creation, merge/release
blocking, ledger persistence). Counts per (module, contract
version); pushbacks 1-3, 4th failure (> 3) rebuilds, 1 rebuild
and 1 targeted round per version, third-failure triage
(structural -> early rebuild, contract/design -> refreeze,
near-complete -> targeted), rebuild-only-implementation with
the old-test first gate, acceptance rebuilds pending user
confirmation, quota blocks consume nothing, cost threshold
escalates rebuilds to the user with full history, corrections
to pass do not unblock, ledger events match the #241 table
with CONDITIONAL mapped to fail at the boundary.
No server wiring, no release endpoint, no task creation, no
DB access in this track; wiring is the next track's job
against this interface.
Single sampling rule; track 3 (#244) is authoritative for WHO is
mandatory, track 5 (#238) for HOW the rest are drawn:
- Mandatory (1.0): golden-task records + rebuild / refreeze /
escalation records + third-failure near-complete still-failed
cases (mandatory_triggers, full list per case). Track 3's old
deterministic every-10th ordinary rule is removed (it drew zero
from batches under 10 ordinary records and was predictable);
track 5's old narrow mandatory list is expanded with golden /
refreeze / escalate. blocked_quota stays non-mandatory.
- Random tiers: normal 0.10, critical-path 0.50, seed required
and recorded, ceil so a non-empty group is never rounded to
zero. 0.10/0.50 are TEST DEFAULTS, not approved production
values; production rates remain a user decision.
- One pass, no double counting: mandatory records leave the
random pools before drawing; a multi-trigger record (e.g.
golden + rebuild) is sampled once with all triggers listed,
and its correction follows the (original, corrected) verdict
pair, never the trigger list.
- cloud/audit_pushback.py::select_audit_sample is now only a
record-shaped adapter over cloud/audit_sampling.py::select_sample
(seed required, retired reason names critical_path-for-kinds
and periodic_sample removed); AUDIT_SAMPLE_EVERY_NTH deleted.
- Correction: original never mutated, correction appended,
corrector != signer enforced in audit_sampling when both names
are supplied; gate effect pinned (correction to pass never
unblocks by itself -- schedule/cancel-and-rerun, blocked until
a new passing re-audit; correction to fail re-blocks at once).
Track 3 / track 5 retrigger vocabularies documented as synonyms.
- Skill text (tightening 3 item 2 + correction paragraph)
rewritten to the unified rule; new test file pins the
boundaries (mandatory union, dedup, ceil incl. the 9-record
case, seed reproducibility, golden-set construction under both
sample vocabularies, adapter order/non-mutation, correction
preservation on both tracks).
Bypass closed: automations._audit_row (direct audit_log INSERT,
best-effort swallow) is deleted. run_due_automation now writes via
cloud.auth.audit_on_conn, the same audit_log + audit_ledger pair
AuthStore.audit() writes, sharing created_at. The audit pair runs
before record_automation_run's commit, so task row, run row and
audit pair commit or roll back together; chain-head IntegrityError
retries the whole block up to 5 times, other errors roll back and
raise (fail-closed). The claim stays committed, so a failed run is
retried next interval, never recorded as a failed run.
AuthStore.audit() now delegates to audit_on_conn; its pair, retry
count and fail-closed semantics are unchanged.
- cloud/attempt_ledger.py: separate task_attempts table (not tasks
columns) with per-attempt rows, MAX+1 attempt_no, append-only
intent; interrupted rows kept but never counted; is_defect is
explicit, never inferred from failure; first defect category
wins; no-ledger tasks stay unknown (None), never backfilled.
SQLite DDL is CREATE TABLE IF NOT EXISTS (idempotent); PG_DDL
is the verbatim manual step for production Postgres, kept out
of POSTGRES_SCHEMA on purpose (migrations coverage untouched).
- cloud/server.py GET /metrics-report: aggregates the ledger
into compute_metrics records; missing ledger table degrades to
unknown rates, not an error. Cost dual-column code untouched.
- cloud/metrics.py: docstring only (ledger is now the source);
compute_metrics logic and all #238 semantics unchanged.
- Tests: 16 ledger tests (store/aggregate/metrics/PG adapter
evidence) + 3 endpoint tests (real rates, old task beside
ledger task, interrupted status non-terminal).
The #241 import resolution deleted the whole 'from cloud.cost import
CostStore, dual_column_view' line to drop the unused alias, which
also removed the CostStore import that _quota_snapshot needs; every
task creation then failed closed with 503 quota_check_unavailable
(4 EndpointChainTestCase failures in test_audit_wiring). Restored
as a plain CostStore import; dual_column_view stays available in
cloud/cost.py for #241's own tests.
- cloud/audit_sampling.py (pure, stdlib): mandatory sampling (1.0)
for rebuild cases and for third-failure near-complete cases whose
continuation round still failed; random sampling for the rest
(normal 0.10, critical-path 0.50, ceil so small groups are never
rounded to zero, seed required and recorded for reproducibility).
- Wrong-verdict handling: correct the original conclusion and
retrigger the corresponding flow (wrong pushback ->
cancel_pushback_and_rerun_audit; wrong rebuild ->
halt_rebuild_and_restore; wrong pass -> reopen_audit_and_pushback;
unknown original -> escalate_to_human); every result, correct or
corrected, is appended to the audit-of-audit record.
- Golden-set approval checklist: list reviewed, criteria defined,
sample exclusion, approver+timestamp -- all human-confirmed before
approval; sample entries (sample=True or sample- ids) are dropped
and counted, never counted as golden.
- metrics.build_report gains an optional sampling_plan section;
without a plan it states sampling is pending, never implying the
audit was re-examined.
- cloud/arch_conformance.py: contract-version (+1 only, approved
change record), forbidden-directory (own dir only, .. and symlink
escapes resolved), forbidden-import layer bans audited against the
baseline (foundation/execution/driver, server sqlite3 Row adapters
deliberately allowed), G2 frozen-stack swap guard; skill-project
one-call check.
- cloud/metrics.py: first-pass/defect/rework rates over terminal
tasks, unknown rates as None (never 0), #212 dual-column per-task
cost (total None when incomplete), pure feedback rules
(tighten/route-with-extra-audit/recalibrate, auto_apply False),
golden-set gate: unapproved sample set is never decision-ready.
- GET /metrics-report (authenticated, member-scoped, same as
/metrics): attempts/defects are None until an attempt ledger
exists, so rates stay unknown rather than fabricated.
- Tests: 29 arch + 15 metrics + 3 endpoint, each check with
positive and negative cases; repo-level scan pins zero layer
violations on the baseline so a future violation fails CI via the
existing host-verify discover run (no workflow change).
Track 3 decisions (audit_pushback) now really block the server merge
endpoint and really land on the track 4 hash chain, via the new
cloud/audit_wiring.py translation layer:
- POST /tasks/{id}/merge ANDs the chain pushback block
(merge_blocked/release_blocked/blocked_by=audit_pushback/round/
failed_checks, replayed per module+contract version) with the
hard gate's five checks in every gate mode; either side blocks
-> 409, status unchanged, audit_pushback_blocked appended to
merge_gate.blocks. Ledger reads that fail (other than a missing
audit_ledger table) fail closed to 409.
- Every pushback/fix/re-audit/rebuild/escalation step is written
to the ledger with track 3 fields authoritative and track 4
storage authoritative; CONDITIONAL maps to fail at the boundary
with the original preserved, history never rewritten.
- Hard-gate 409 blocks codes feed pushback packages machine-first
(branch protection survives as its own extra check).
- Re-audit QA guard: a claimed pass with un-rerun or failed QA
links never reaches the chain.
- Track 4 compat: one audit_wiring_event action for escalation /
blocked-quota / audit-sample kinds, counted in the timeline.
- Resolved tension (documented in audit_wiring.py): a meta-review
correction to fail re-blocks at once, but a correction to pass
does NOT unblock by itself -- track 3 schedules a re-audit,
it does not release the gate; track 4 failure counts still use
effective verdicts, so the two views may disagree while a
re-audit is pending.
Tests: cloud/tests/test_audit_wiring.py (18 items), incl. a real
HTTP end-to-end chain (fail -> 409 -> fix -> re-audit pass ->
200) and a separate QA-link chain test.
Independent audit (no code, independent re-run, separate signature;
CONDITIONAL never passes; deviation needs due todo) and automatic
pushback (fail -> derived fix subtask, merge/release blocked,
package with per-item evidence, full re-audit, 3-round limit with
a separate CI auto-fix counter, ledger hash-chain interface for
track 4, gate-block interface for track 1). Over 3 failures per
module per contract version: rebuild only that module from the
frozen contract (old implementation archived, avoidance list),
contract/design root cause -> G2 re-freeze escalation instead.
Tightenings: 3rd-failure triage (structural early rebuild /
early refreeze, near-complete one targeted round), rebuild keeps
tests and passed assets with an old-test first gate, audit-of-
audit sampling/correction and G6 user-only acceptance plus
acceptance-rebuild user confirmation. QA full chain: self-test
three parts, PRD coverage, agent six, cross-client matrix with
honest uncompiled verdicts, G5 gates, impact-scoped regression,
dual QA/audit signatures, failed-link re-runs in re-audit.
Skill text is purely additive (+459/-0) against main v0.2 text;
port this section into v0.3/v0.4/v0.5 texts on merge, never
overwrite them with this file.
Integration arbitration, not a PR commit: v0.3/v0.4/v0.5 synced body
changes into appendix chapters 00/02/04/06/07 while templates/roles/
sources stayed at v0.2, so check_appendix failed 6 chapters; and
09-ops/10-iteration chapters (v0.5) were invisible to the strict
chapter regex because of the ', v0.5 new' title suffix, breaking
the #214 matrix-vs-appendix test. Direction is sources follow the
appendix (appendix is the tenant-visible authoritative version;
never overwrite v0.5 text with v0.2). 09/10 headers normalized to
strict form (suffix only); 09/10 sources split mechanically from
the appendix; EXPECTED_IDS extended to 11 chapters in the checker
and both appendix tests.
- test_skill_appendix_sync: all 9 appendix chapters 00..08 in order,
body role-matrix set == appendix role-title set, authority
statement on both sides, GitHub guarantee in body + named
appendix chapters 00/04/07, seeded copy byte-identical with
full appendix. Previously only 00/08 existence was asserted.
- test_skill_seed_pinning: only SKILL.md copied (pack extras never
land, never resurrected), tenant extras left alone, tenant-modified
SKILL.md never overwritten across repeats, identical = noop,
deleted SKILL.md re-seeded (documented boundary, not deletion
preservation).
- skill_seed.py: evaluation note only (why whole-dir copy is not
done); no code change.
- CostStore: per-model breakdown, report_by_creator (JOIN tasks.created_by),
dual-column reports with priced_usd_micros, unpriced_models listed
separately and a billing note (proxy price, not the relay bill);
_report itself untouched, old exact-key tests keep passing.
- Hard cutoff: cost.quota_cutoff_reason is the single USD/unpriced-cap
decision; worker._quota_exceeded now delegates to it (identical
messages); claim gate, turn boundary and poll kill stop spend and
record interrupted + quota reason + audit; auto-resume re-checks
live quota before requeueing.
- Approval timeout (G7 ops metric): cloud/approval_timeout.py —
default stays 600s, >20% timeout rate recommends 1800s, <5% keeps,
mid-range keeps; stats + recommendation + info log; wired into
/metrics (member projects, no log spam) and a global-admin
/admin/ops/approval-timeout endpoint (sweeps overdue first).
- Server: task/project cost endpoints return dual-column reports;
new GET /users/me/cost and GET /admin/cost-by-creator.
- Tests: cloud/tests/test_cost_quota.py, 10 tests.
Core artifacts travel as an explicit manifest; coordinator triple-checks
(core report vs manifest vs recomputed hash) before cleanup, regression
rewrites files verbatim, recomputes sha256, stops on ARTIFACT-HASH-MISMATCH
and only then runs the same core code plus import-only regression tests.
GET /full-auto/status, GET /full-auto/config-plan, POST
/full-auto/validate-action|merge-gate|ci-fix|release-plan|
rollback-plan, POST /full-auto/execute (global admin only,
FULL_AUTO_ENABLED + verified installation + branch protection
verified via GitHub API, else 403 + audit), GET /full-auto/audit.
Guards reused from cloud/full_auto.py; token only in Authorization
header; audit via audit_record redaction into AuthStore audit_log.
Only pure computation/validation sub-blocks of _handle_github_webhook,
_create_followup, _deliver_to_thread and _task_chat_history are extracted;
all DB writes, audits, external calls and side-effect blocks stay in the
callers, error strings/codes and response shapes verbatim. _deliver_to_thread
uses the shared goal classifier only for /goal input: its non-goal terminal
path still goes to CAS revive (not 409) and its merged path keeps its own
wording. New tests: test_longfunc_remaining.py (15).
_continue_task 226->168 via _validate_continue_options;
_task_wait 158->139 via _validate_wait_params;
_task_thread_read 187->152 via _summarize_trace_entries;
_metrics 123->104 via _summarize_metrics_tasks.
No behaviour, message, status or route change; error strings verbatim.
New tests: test_longfunc_extra (9). Cloud full suite: 1439 OK, skipped=4.
Follows #229 pattern; must merge after #229 and be rebased/checked then.
Move the /tasks/* sub-route out of _dispatch into _dispatch_task_route
(query passed explicitly; the files/tree/thread branches use it), move
the push-subscription DELETE branch out of _dispatch_project into
_project_push_subscription_delete, and turn the goal-validation block
of _create_task into the pure module function
_validate_task_goal_options (returns error string instead of sending
400, caller sends it unchanged). Statements are moved verbatim except
indentation and the goal send+return -> return-error transform; no
behaviour change. server.py module split intentionally not done: it
remains a larger, higher-risk refactor.
Add test_longfunc_split covering the pure goal validator's success
paths and all eight error strings.
- ensure_workers_table: new tables include host TEXT, pid BIGINT,
address TEXT; old tables gain them via _ensure_hostpid_columns
(PRAGMA column check, missing-only ALTER, old rows keep NULL,
never fabricates host/pid). SQLite and Postgres share this path
(PG via the db adapter's PRAGMA translation); POSTGRES_SCHEMA
deliberately still excludes workers, which is volatile heartbeat
data in UNMIGRATED_TABLES (adding it broke the P2 migration
alignment tests, verified and reverted).
- heartbeat/register: given host/pid/address refresh, omitted
values preserve the old row; address has no derived source.
Local worker passes host_name + own pid; HTTP heartbeat only
stores caller-supplied values, never the server's own.
- classify_stale_workers: kill only when age strictly exceeds the
threshold AND host matches local host AND pid is proven dead;
equal threshold, foreign host, missing pid, or live/unknown pid
are list_only. Classification only, never kills, never matches
by name/prefix, never touches infra containers.
SKILL.md header pointed at same-dir README.md but the pack had no
such file (CHANGELOG was supplied by #217, README was not). This
track adds the README only: purpose, chapter map, appendix
provenance entry point, seeding behaviour, and known boundaries.
It states no rule, changes no SKILL.md semantics, and does not
duplicate the CHANGELOG/script explanations -- overlapping parts
are pointers, with an explicit division-of-labour section.
Deletion retention is written only as the tombstone-design
placeholder, deferring to the skill_seed implementation, because
no tombstone implementation has landed. Test pins that every
file SKILL.md cites as same-directory exists (CHANGELOG, README).
SKILL.md claimed its appendix was spliced verbatim from
templates/roles/ by a generation script (see CHANGELOG), but the pack
contained none of the three. Add the nine role sources (mechanically
split from the v0.2 appendix; SKILL.md untouched, zero semantic
change), tools/check_appendix.py (verify + --emit, stdlib only, never
edits SKILL.md), CHANGELOG.md, and test_skill_appendix_provenance.py
(5 tests). Seeding unchanged: skill_seed copies SKILL.md only, so
templates/tools/CHANGELOG never reach the tenant home.
- cloud/merge_gate_hard.py (new): pure hard gate; mode resolution is
env-only (unset/invalid -> hard, fail-closed); five checks plus
branch protection must be protected; unknown/unprotected block
even with explicit confirm; no-evidence blocks (fail-closed);
directory check reuses #209 check_directory_isolation (v0.3 rule);
blocks codes + merge_gate_blocked audit detail are the track-3
pushback interface (no pushback state machine in this track).
- cloud/server.py: POST /tasks/{id}/merge uses hard mode by default;
record/enforce/off only when explicitly set via env (preserving
#209 semantics); body merge_gate_mode is ignored server-side so
callers cannot downgrade the gate; every gate block writes a
merge_gate_blocked audit before returning 409, status unchanged;
hard-mode body type errors return 400.
- docs/merge-gate-branch-protection.md (new): protection payload
(PR + strict status checks + no force-push + no deletion +
enforce_admins), empty required_checks = unconfigured, Pro/public
prerequisites, 403 causes; states protection is NOT claimed enabled.
- SKILL.md (gate-mechanism wording only): four checks -> five checks;
body + coordinator + client appendix gate paragraphs updated to
default hard, merge_gate_blocked audit, payload doc reference.
- tests: 31 new hard-gate tests (pure + endpoint + skill text);
#209 endpoint tests now set record env explicitly (their intent);
test_server_merge 200-path calls now carry full hard evidence.
- cloud/merge_gate.py: pure gate; directory check reuses the v0.3
git diff --name-only merge-base command; protection modes
record (default, explicit confirm required when unprotected),
enforce (always blocks when unprotected), off (explicit only).
Confirm covers protection risk only, never directory violations.
Protection status is caller/env-declared only; unknown is a
recorded risk, not a claim of protection.
- server POST /tasks/{id}/merge: applies the gate only when gate
evidence is supplied (or env declares protection); no-evidence
calls behave as before plus a merge_gate field in the response.
Blocked merges return 409 without changing status; confirmed
unprotected merges write a merge_gate_risk audit entry.
- skill_seed.py: whole-directory seeding evaluated and rejected
(cannot preserve tenant deletions; no benefit while the pack
holds only SKILL.md); SKILL.md-only behaviour unchanged.
- software-engineering SKILL.md: gate paragraph added to the body
and to the coordinator + client appendix sections (v0.2 text on
this base; port to v0.3 text on rebase after #199 merges).
- tests: 16 new tests covering gate modes, directory reuse,
body/appendix bidirectional assertions naming the missing side,
SKILL.md-only seeding and tenant-edit preservation.
Stacked on v0.3 + v0.4 (HEAD d6522ba feat-full-auto): v0.3/v0.4
content preserved, not reverted. Adds G7 post-release ops and
observation (monitoring, errors, cost/quota, feedback, hotfix loop,
rollback triggers, ops manual) and G8 iteration and scaling (v1.1
handover list, contract v1->v2 protocol, new-end onboarding,
scale/performance, deprecation), with G6->G7 and G7->G8 gates,
DAG, gate table, coordinator derivation, G0-G8 todos, checklist,
09-ops and 10-iteration role templates, and ops-manual / v1.1
handover templates in both body and appendix. Settles web+React+
multi-viewport focus, cross-viewport E2E, web single-end release
with iOS/Android native postposed, open-source-first, and the
TypeScript full-stack default (user-request-only language change;
codex-app-server itself stays Python). Pending #199 and #205
merge, then one rebase round. PR only, do not merge or deploy.
- P1-1 directory-isolation executable check: git diff --name-only
against merge-base, output recorded in merge record
- P1-2 derivation checklist table (8 required items, missing one blocks dispatch)
- P1-3 QA split into single-client review and cross-client E2E with
fixed deliverables and agent-test attribution
- P1-4 sub-task slim-pack trimming rules (listed chapters only, never full copies)
- P2: inter-thread Q&A default policy, docs/decisions.md append-only,
per-client onboarding docs (web first at G4a), v1.1 handover checklist
- Body and appendix kept in sync; new lightweight keyword test
- worker._tenant_dirs: playwright refresh log used undefined 'task';
use the method argument project_id (ruff F821, real bug: NameError
aborted tenant seeding when a playwright script was refreshed).
Regression: cloud/tests/test_worker_playwright_seed_f821.py (2 tests;
pre-fix run reproduces NameError at worker.py:2780).
- harness2.assign_ids: hashlib.sha1(..., usedforsecurity=False) -- the
digest is only a deterministic shuffle seed, not a security use;
digest value unchanged, bandit B324 no longer fires (High 5 -> 4).
- harness2.select_final: rename ambiguous loop var 'l' to 'lab'
(ruff E741, behaviour-invariant).
- Deliberately NOT changed: host.py shell=True x4 (see PR body -- all
four are intentional shell commands, changing semantics is unsafe),
remaining ruff 109 / mypy 24, workflow thresholds (75% coverage
gate unchanged; no gate was relaxed or removed).
Phase 1 of audit-hardening CI expansion (baseline origin/main aebc623):
- lint: ruff E9 hard gate; full E,F@120 report-only (baseline 111 items)
- typecheck: mypy default mode, production scope, report-only (baseline 24)
- security: bandit production scope, report-only (baseline 287, High 5)
- depscan: pip-audit on requirements-ci.txt, hard gate (0 vulns; repo has
no production dependency manifest, see summary)
- coverage: production 58 files, fail-under=75 (baseline measured 76%)
- release build: E9 + pip-audit hard gates before packaging; full reports
stay in host-verify to avoid drift
No production code changed; CI billing failure mode unchanged.
Goal tasks pushed back by the audit loop lost the packet preamble on
the next round: claim-time _apply_audit_loop_packet pasted it into the
in-memory prompt/goal, but the next round collided with the existing
host task name, fell back to breakpoint goal-resume (no --goal-text),
and the host reused the stored goal, discarding the preamble.
Fix (minimal, direction (c) trimmed): host goal-resume gains
--resume-note, accepted only by the breakpoint branch and prepended
to the first turn prompt of this resume only -- never written back
to the stored goal, never touching rounds or verification. The worker
stashes the same preamble as the memory-only _audit_loop_resume_note
at claim time and forwards it on breakpoint resume (Local/Docker
goal_resume_task and the popen goal-resume command, each guarded to
ignore the note when goal_text is set for a new-goal resume).
Accounting untouched: audit 3-round cap, goal_max_rounds/quota, and
DB goal columns are unchanged; non-goal pushback is unchanged.
Tests: cloud/tests/test_audit_n12.py (9 tests) -- host first-turn-only
note + stored-goal/rounds, worker exists-fallback end-to-end with
--resume-note, non-goal regression, goal-column accounting, and
Local/Docker/popen command shapes. Full cloud suite: 1760 OK
(skipped=11), baseline 1751 + 9 new; tests/test_host.py: 62 OK.
- I-3: new models.evaluate_capability returning a CapabilityVerdict
(ok/missing/unknown from check, known/unknown_missing from split,
status/verb derivations); decide_routing, worker gateway check,
worker persisted recheck, dry-run tier+gateway checks and server
shadow endpoint all call it. Blocking rules stay at the call
sites (gateway has no worker-default fallback, recheck blocks
only on known, dry-run/shadow never block); all user-visible
audit details, dry-run notes and shadow notes are byte-identical.
- I-4: new models.merge_policy_documents shared by
load_routing_policy and dry_run_from_env; the old tiers-update
block is kept verbatim (it was a no-op after update(): JSON
tiers replace the whole map, file tiers absent from JSON are
dropped -- preserved, not fixed, as that would be a behaviour
change). dropped_tiers/partial_failures remain load-time
records, not merge products.
- Tests: test_audit_batch3.py, 13 tests (verdict matrix + parity
with check/split, decide/dry-run wording, merge boundaries +
300 seeded random pairs + loader/dry-run agreement).
Full suite: 1694 OK (skipped=11), baseline 1681 + 13, 0 failures.
- I-1: evaluate_shadow counts dropped rows by cause (unparseable /
unknown_tier / explicit / duplicate) with the identity
audit_rows_total == routed_task_n + dropped_total; the shadow
endpoint passes created_at/id through and reports audit totals
next to sample counts.
- I-2 (option b): parse logs a warning naming each tier dropped for
an invalid mode/requires; load attaches partial_failures to the
policy; the worker records a task-level
worker_routing_policy_load_failed audit for partial failures
(same action as full failure, detail says failure=partial);
the shadow report carries the policy fingerprint, dropped tiers
and partial failures (no task context on that path, so no
task-level audit there).
- I-5: coordinator gate audit lookup failure now raises
_CoordGateLookupFailed and the merge converts it to escalate
(no requeue, no round consumed); persisted-model lookup failure
and required-tools check failure each write a named audit row
before falling through to dispatch/allow; audit-loop evaluate
failure stays escalate (regression test).
- Test fixtures: coordinator suites build audit_log via AuthStore
as production's first real audit does, since their stubbed
_audit never creates the table.
- B-A: migrations.py tasks backfill loop now types audit_loop as
INTEGER (other 6 columns stay TEXT), matching db.py backfill,
server/worker ensure lists; TEXT readback '0' is truthy and
would silently enable the audit loop for audit_loop=0 rows.
- B-C: gateway preparation failure early-return now calls
_running_clear(task_id) next to _cleanup_task_seed, matching
missing-key/repo/capability/tool gates; previously the capacity
slot leaked until worker exit (default capacity 1 blocks reclaim).
- Tests: cloud/tests/test_audit_batch1.py (3 tests): migration
ALTER types, INTEGER 0 readback not enabled via real worker
row mapping, gateway failure clears slot and capacity reclaim.
_claim doc 'three gates' list left unchanged (full early-return
gate enumeration not verified this batch; residual wording).
- No other audit items touched (I-1..I-14, N-1..N-12).
- M1: test_goal_branch_via_convert_queued_then_claim actually calls
server._goal_convert_queued on a queued auto-resumed task before
claiming, instead of only inserting the post-conversion state.
- M5: test_worker_routing_audit_rows_carry_no_fingerprint drives a
real run_one with an active policy and asserts worker_routing
audit rows exist and carry no fingerprint (negative control
verified: injecting fingerprint into the detail fails it).
- M2: 'all callers' wording tightened to the real counts
(load_routing_policy: 2 production call sites; evaluate_shadow: 1).
- M4: content_fingerprint docstring now states the fingerprint covers
the full normalized payload including fallback_models, which the
router deliberately does not apply (fail-safe direction).
No behaviour change; M3 DESIGN.md edits are outside the repo.
- db: translate INSERT OR IGNORE ... SELECT to ON CONFLICT DO NOTHING
(VALUES-only regex left SELECT form untranslated -> PG syntax error);
register task_audit_loop_state (task_id) in _OR_REPLACE_PK so PG
INSERT OR REPLACE becomes a real upsert instead of DO NOTHING.
- server: _backfill_default_quotas uses SAVEPOINT containment and no
longer commits the caller's transaction, so a backfill failure cannot
roll back init_db's migrate ALTERs.
- worker: _coord_child_rows rolls back before column-fallback retries so
a missing-column failure on PG does not cascade into
InFailedSqlTransaction.
- tests: test_pg_init_fix.py (translation, SQLite backfill/upsert,
real-PG fresh-init + merge + upsert + backfill containment, skipped
without CLOUD_TEST_PG_DSN/driver).
Parent audit loop sits after all children are aggregated and
before the parent final state; the single-task _audit_loop_wire
is never applied to parents or children. Parent gates are
per-child gates child:<name> (completed only); execution
failures map to one deterministic finding per child; subjective
findings are split off before evaluate and only audited as
needs_review. Gate-rejected children (capability/tool/approval
audit rows, or distinctive error markers) bypass evaluate and
escalate directly without consuming a round; tolerated, paused
and quota-missing parts are excluded, never findings. Rounds
are counted only in the parent task_audit_loop_state; on
pushback the packet is stored on the parent and split per failed
child into that child's audit_loop_packet column, consumed once
by the existing packet-injection path at next claim; pass and
escalate clear parent and child packets. POST /tasks now allows
audit_loop on coordinator parents (the old 400 rationale is
gone); children still take no external switch.
Audit-loop wiring, worker completion path only: after the final
write-back (and only when the final is completed/failed, not quota,
not cancelled, not goal-converted, and the task carries audit_loop),
collect this attempt's deterministic gate results (execution,
goal_verify only when host reported goal_met, capability /
required_tools / gateway as passed-or-inapplicable markers, since a
pre-dispatch gate rejection early-returns before any runner round
and requeueing it would deterministically re-fail the same gate)
and call audit_loop_evaluate. Pushback stores the packet JSON in
tasks.audit_loop_packet and CAS-requeues the task; the next claim
prepends the packet preamble to prompt (and goal) with a data-only
separator, DB prompt untouched. Pass clears the packet. Escalate
sets failed (VALID_STATUSES has no needs_review, never invented)
with an escalation error and clears the packet; subjective
needs_review leaves the outcome untouched and only audits.
New tasks.audit_loop / audit_loop_packet columns follow the
attempts DDL pattern in worker ensure, server migrate, PG schema,
PG backfill, migrations backfill, and POST /tasks accepts
audit_loop (True/1 only, coordinator tasks rejected like v9).
Coordinator parent tasks return via _run_coordinator before the
attempt loop and coordinator children never inherit audit_loop,
so coordinator wiring is left as legacy, unchanged.
Audit loop engine (cloud/audit_loop.py): Finding/PushbackPacket/LoopState pure functions; deterministic gates only, subjective -> needs_review; packet completeness enforced; standards-weakening -> escalate; regression and missing-gate checks enforce full re-run; MAX_PUSHBACKS=3, round>=3 -> escalate. Persistence in new single-row task_audit_loop_state (attempts DDL pattern: module CREATE_TABLE_SQL + db.py POSTGRES_SCHEMA + migrations list + server init + worker ensure); task_attempts NOT reused (attempt_no PK collision) and task_followups NOT reused (user messages). Worker.audit_loop_evaluate is the explicit opt-in hook (task audit_loop field only, no env switch, never called by run_one); escalate audits worker_audit_loop_escalated; corrupted/missing count fail-closes to escalate.
Parse: agent_tools/tool_approvals column parsing unified into agent_def shared helpers (parse None = undeclared; server keeps None, worker maps to []/{} at _task_from_row, dispatch behaviour unchanged). Goal: _apply_resume_preamble goal branch proven reachable via _goal_convert_queued on a queued auto-resumed task, kept with a test and corrected docs (not dead code). Fail-open: audit write failure and preamble lookup failure now log warnings (best-effort unchanged); partial policy failure and shadow parse drops documented as known gaps. M7: shadow endpoint explicit_capability_failures_truncated test. Fingerprint: RoutingPolicy.content_fingerprint (canonical JSON sha256, version excluded) surfaced only in dry-run output, never in audit rows. M1 wording: legacy gap stated as fixed stock persisting per-task until completion, not converging.
N1-1: writeback scope verified too broad (any chosen, incl. shadow
applied=False, worker_default and worker_explicit) and narrowed to
source=policy + applied only; writeback now records a
worker_model_persisted audit row as the identity marker.
N1: persisted models are rechecked against current tier requires
before dispatch on later attempts (fixes4 split known/unknown):
active + known-missing fails fast (worker_capability_recheck_
rejected, no runner), unknown-only and shadow only audit.
User-explicit, worker-CLI-explicit and gateway-filled tasks skip
the recheck; legacy models are recognised by a policy applied
worker_routing row. Recheck never rewrites tasks.model and never
re-decides.
Explicit gateway capability now splits known/unknown: known-missing
explicit models fail-fast under active (same shape as the gateway
default gate), unknown explicit models stay audit-only, shadow never
blocks; explicit priority now protects which model the user chose,
not that the chosen model must run. Shadow report reads
worker_capability_gateway_explicit rows as explicit_capability_failures
with a field note distinguishing the three explicit counts.
N2: coordinator requeue CAS rowcount=0 branch covered via a proxy
connection test (no refactor needed). N3: shadow endpoint test on
the create_server + real HTTP skeleton (capability, truncated, and
explicit-failure fields). N4: creation-layer rejection tests for
tools/instructions/tool_approvals on goal/coordinator tasks.
N5: model writeback failure logs a warning (best-effort kept).
N1: writeback capability-recheck trade-off documented in comment
only; recheck itself is a later batch.
B1: gateway capability check now triggers on walking the sub2api
gateway, not on the gateway having filled a model; task-level
explicit models are audited (worker_capability_gateway_explicit)
but never blocked, preserving explicit priority; gateway-default
models still fail-fast under active.
I1: dry-run adds a gateway-model section, sourced only from
CLOUD_SUB2API_DEFAULT_MODEL (never queries the relay); an active
tier whose requires the gateway model fails or cannot verify
blocks, shadow only warns.
I2: shadow review gate now counts completed samples
(completed >= 20), same basis for sample_shortfall; the endpoint
attaches per-tier capability status (read-only policy load) and
flags capability-failing tiers in the recommendation.
I3: configured-but-broken routing policy logs a warning in
load_routing_policy (unconfigured stays silent); the worker sets
routing_policy_load_failed and audits
worker_routing_policy_load_failed per task.
I4: resume preamble wraps prior output/error as data with an
explicit do-not-execute constraint; position verified in run_one
(tree -> preamble -> project -> original) and the docs corrected
from the false 'closest to the original prompt' claim.
I5: decided model is written back with a CAS UPDATE
(model IS NULL OR model=''), so later attempts reuse it;
explicit models are never overwritten.
I6: agent_* columns added to worker._ensure_result_columns and
db.init_postgres_schema backfill; migrations backfills them for
tasks before insert (sender_task_id pattern); column-level tests
for agent_tool_approvals and task_attempts.
I7: coordinator requeue adds AND status=<snapshot> (rowcount 0
skips requeue/audit) and filters paused_by_user / user_cancel /
user_pause children, which are never revived or cleared.
I8: explicit-model tasks (including coordinator children that
inherited a parent model) audit a source=explicit worker_routing
row; evaluate_shadow counts it as per-tier explicit_n only,
never routed_n, never a dedup slot.
M1/M2/M3/M5/M6/M8/M10: dedup slot moved after tier validation,
shadow sort doc fixed, truncated recommendation notes truncation,
seed cleanup on gateway/quota/goal-cap early returns, tools-only
coordinator definitions rejected, preamble goal branch marked
effectively unreachable, DESIGN.md drift corrected (outside repo).
M4/M7/M9: documentation/new-field only (pass_rate_fully_measured,
whitespace-model shadow-drop note, bound-note race note); old
field semantics unchanged.
- RoutingPolicy.fallback_models: move the cleaning loop into
__init__ (it was unreachable after tier_mode's return, so the
attribute never existed); still parsed but not applied.
- Capability fallback: RoutingDecision gains capability_fallback;
the worker audits worker_capability_fallback when it fires, so
a tier-model rejection followed by a worker-default fallback
is no longer invisible (source stays worker_default, no
model_routing event, so shadow samples are unaffected).
- Registry: resolve_registry_agent checks skills directly
against the loaded definition instead of re-parsing a
hand-built dict, so future fields cannot be silently dropped
the way tool_approvals was in v9.
- Shadow: evaluate_shadow sorts rows earliest-first by
created_at (id tiebreak) when timestamps are present, so
dedup no longer depends on caller row order; the server
path (which strips created_at) is unchanged.
- Gateway: when sub2api fills the default model, the worker
now checks it against the tier's capability requirements
with the same mode semantics as v8 -- shadow only audits,
active fails fast with worker_capability_gate_rejected and
never calls the runner. No worker-default fallback here:
the gateway model is the task's gateway model, falling back
to a worker default would run against the wrong gateway.
Agent definitions gain tool_approvals: known tool -> ask/auto/deny.
Unknown tool/policy is a 400, and required+deny is a 400
contradiction. The declaration is persisted (tasks.agent_tool_
approvals, PG + migrate), echoed in GET and the creation response
(via to_summary), and resolved for registry definitions -- fixing
a latent drop where resolve_registry_agent re-parsed without new
fields. At dispatch the worker audits the declaration and fails
fast on a defensive required+deny re-check (v3 tools-gate shape,
runner never called); goal/coordinator tasks reject the field at
creation like instructions. effective_tool_approvals maps a
declaration + host global approval to the actual host answer,
with ask_honoured always False: host answers elicitation with
one global switch and carries no tool identity, and the approvals
queue is record-only with no runtime ask channel, so per-tool
interception and credential scope are explicitly NOT claimed.
Policy carries an operator-declared model_capabilities directory
and per-tier requires (tool_calling/reasoning/vision booleans,
min_context_tokens). decide_routing checks the policy tier model
in active mode: capable applies, incapable falls back to a capable
worker default, and if both fail or are unknown it returns
capability_blocked, which the worker turns into a failed task
(v3 tools-gate shape, runner never called). Shadow mode only
annotates the reason, never blocks. Explicit task and worker CLI
models bypass the check by design. Invalid requires drops only
that tier (fail-closed, like invalid tier mode). Dry-run reports
per-tier capability status and blocks on active tiers that fail
or cannot be verified.
Production code had no golden tasks, scorer, or rubric
(harness2.py states this explicitly); routing shadow gives
cost/status only and admits counterfactual not_measured,
so 'is the suggested model actually better' had no
mechanism. This track adds a pure module + CLI that scores
already-completed task records against human-curated golden
tasks and compares two runs, without running tasks, spending
model cost, touching the DB, or activating anything.
Golden tasks: JSON {id, prompt, tier, checks}; only four
deterministic check types (contains, not_contains,
status_completed, max_tokens, max_cost_usd, the last in
cost.py microdollar units); invalid definitions fail loudly.
Scoring: per-task score = passed / measured checks;
cost checks with missing/unpriced data are unknown, excluded
from the denominator and reported separately (never counted
as pass or fail); empty result_text is a string-check fail,
not unknown. all_pass means 'all measured checks pass' and
is reported alongside fully_measured_n so a pass rate is
never misread as 'all checks passed'. Comparison: per-run
stats, paired delta over golden tasks present in BOTH runs
(unpaired listed separately, missing listed, unmatched
records excluded with a warning), duplicate task_id rejected.
Paired n < 20 (same threshold as shadow) => ready_for_
decision false, no significance or p-values computed.
Deliberately NOT included: a task runner (golden tasks
still need a separate run-and-record step or manual task
creation), an LLM judge or quality score, and any routing
activation suggestion. A contains check can be gamed by
echoing, so golden sets must be human-curated with checks
that require actually doing the work; the example file in
eval/ is format-only, not a decision-ready set. Tests:
test_eval_loop.py (8 tests).
Foundry 'platform restarts with original input' gap, after
reading the existing mechanisms: creation idempotency
(UNIQUE project_id+idempotency_key), goal checkpoints
(task_checkpoints per round, same-thread breakpoint resume
with no-progress fuse), merge/cost/notification CAS, and
coordinator child reuse already exist. The uncovered segment
is non-goal platform-interrupted tasks: auto_resume requeues
them fresh in the same worktree after clearing result/error/
interrupted_reason, so the new run gets only the original
prompt and cannot tell what the previous attempt had done.
This track adds, in worker/server layer only (no host.py or
codex thread change): task_attempts ledger, recorded by
auto_resume_interrupted in the same transaction before
clearing (attempt_no = old auto_resume_count + 1, kind
goal_breakpoint|platform_fresh, interrupted_reason,
checkpoint_round, result/error truncated to 2000 chars,
INSERT OR IGNORE keeps the first record on duplicate);
resume preamble injected in-memory only at claim for
platform_fresh reruns (previous attempt facts, truncated to
1000 chars, plus check-worktree-first and verify-before-
recreating-external-artifacts instructions; skipped for
coordinator/_resume/breakpoint tasks and when no record
exists, never fabricated); attempts surfaced read-only in
GET /tasks/{id}/timeline under existing project-read
permission. Fresh PG schema, migrations table list, and
server/worker ensure paths share one DDL.
Deliberately NOT claimed: the preamble is an instruction,
not a hard guarantee like creation idempotency or merge
CAS; the ledger does not track worktree files or external
artifacts (no worktree snapshot, executor-dependent paths;
hard dedup for those still rests on their existing gates);
manual retry paths do not go through auto_resume, so they
get no ledger record or preamble. Tests:
test_attempts.py (7 tests).
host