feat: add Lotus Family auditor v0.1

[safal207]

What changed?

Implements the first executable Lotus Family conformance slice from #230 and extends it with a causal spacetime system model.

Current exact head: 95d0c15a6a52ac84224ff830ba679b624133bfae

Current exact tree: bfa56e436f8af2fae2577c29e8a984fddedffd59

The auditor evaluates supplied materialized snapshots and returns PASS, DRIFT, or UNKNOWN while remaining read-only and audit_only.

Current model:

• 49 system nodes;
• 64 unique directed relationships;
• 36 connected routes across causal, spatial, temporal, hierarchical, and trajectory views;
• bounded snapshot PASS separated from independently verified exact-head freshness;
• route-to-graph identity binding through graph_id;
• full authority-boundary assertions for ownership, approval, execution, delivery, and merge.

The final fail-closed policy rejects or isolates non-executing/path-like shells, custom shell templates without explicit fail-fast behavior, direct or quoted eval terminators, heredoc payloads, directory mutation, cyclic needs, arbitrary setup commands, commands after unproven predecessors, cross-workflow state leakage, empty compact model collections, step-level and job-level non-gating continue-on-error, and default pytest discovery influenced by unproven repository configuration. Action prerequisites require an explicit manifest trust entry and a full immutable SHA; that identity trust remains distinct from proof of action behavior. Explicit contains_any discovery now binds every parsed command target to one checked test_path; every executed Python or Elixir contract-test source must match its manifest-pinned SHA-256; repository-local pytest, pytest-runtime-dependency, bytecode, native-extension, and startup-hook shadows fail closed. Manual-only dispatch cannot establish CI evidence, and job-level containers are rejected because their executable provenance is not proven.

Type of change

• Documentation
• Gate or verdict semantics
• Evidence artifact or verifier
• Authorization / credential / environment handling
• Recovery or escalation behavior
• Tests / CI / tooling

Why it matters

A workflow must not receive a false conformance PASS merely because test-command text appears somewhere in YAML, runs under a shell that can mask failure, belongs to a job whose failure is ignored, or is silently narrowed by repository pytest configuration. The public policy requires a provably executable, reachable, failure-gating regression-test path while preserving uncertainty about caller-claimed repository provenance and leaving every consequential action with an explicitly authorized human.

Exact-head validation

Exact PR head SHA validated:

95d0c15a6a52ac84224ff830ba679b624133bfae

Validation command:

python -m unittest discover \
  -s standards/lotus-family/conformance \
  -p 'test_*.py' \
  -v

• Validation was run or rerun after the most recent PR head change.
• Evidence, screenshots, and expected output apply to the exact SHA above.

GitHub Actions and a fresh Codex review were requested for this exact head and are pending. Local exact-tree replay completed 112/112 conformance tests successfully; JSON parsing, compileall, and diff checks passed. Elixir/Mix is unavailable locally, so hosted CI remains authoritative for that gate. Older-head hosted evidence is stale; merge remains blocked until the exact-head checks and fresh review complete without actionable blockers.

Lotus judgment check 🌸

Does this judgment show its evidence, preserve uncertainty, and leave consequential action under explicit authority?

• A positive PASS is supported by configured contract evidence and executable CI discovery, not merely the absence of a blocking signal.
• Missing, stale, conflicting, inaccessible, or provenance-unverified evidence remains visible and is not converted into confidence.
• The verdict does not execute an external action or manufacture ownership, approval, delivery, deployment, or merge authority.
• Stop reasons, routes, hashes, configuration evidence, and evidence traces remain stable, replayable, and open to human challenge.

Lotus note — one concrete sentence:

A supplied snapshot may receive a bounded conformance PASS, but it retains limitation.identity_unverified until a trusted materializer independently binds the bytes to the claimed repository and commit.

Compatibility and authority

• Preserves the manifest schema and PASS / DRIFT / UNKNOWN result contract.
• Preserves the causal graph and route artifacts as a compatible lower-level layer.
• Keeps the public workflow API as the final fail-closed policy boundary.
• Hashes known pytest configuration files and rejects active default-discovery scopes rather than silently trusting them.
• Pins and verifies every configured executed Python or Elixir contract-test source without widening snapshot identity or execution claims.
• Does not fetch remote repositories or claim trusted materialization in v0.1.
• Grants no ownership, approval, execution, delivery, deployment, or merge authority.

PASS != APPROVED != MERGED.

Evidence

Exact-head behavior covered by executable regression tests includes:

• independent parsing of multiple workflow documents;
• no cross-workflow PYTEST_* leakage;
• fail-fast reachability inside a script and across ordered run steps;
• custom bash {0} / sh {0} templates rejected unless explicit -e is present;
• step-level and job-level continue-on-error rejected for contract gates;
• expression-driven job failure policy rejected as unproven;
• workflow evidence requires a literal push, pull_request, or merge_group trigger; manual-only dispatch is insufficient;
• job-level container execution is rejected rather than trusting shadowable Python, Mix, or Elixir binaries;
• malformed hybrid uses + run steps rejected as non-executable evidence;
• only manifest-trusted, full-SHA action prerequisites accepted; mutable refs remain blocked;
• direct workspace-writing setup cannot establish a later test gate;
• the configured Pythia contract runs directly under Elixir before Mix bootstrap and does not evaluate mix.exs;
• each explicit discovery pattern targets only its declared, checked test_path;
• each executed contract-test source is verified byte-for-byte against its manifest pin, so phrase-preserving no-op or early-exit replacements produce DRIFT;
• Mix-default adapters accept only a closed literal project subset; executable values, arbitrary test-helper code, and every Code.require_file dependency fail closed;
• implicit shell analysis limited to literal POSIX runners;
• an explicitly ignored earlier failure may permit a later gate step;
• known pytest configuration files hashed as evidence;
• repository-local pytest and pytest-runtime-dependency source/packages (including the bundled py shim), sourceless bytecode, native extensions, and Python startup hooks rejected before discovery;
• active [tool.pytest.ini_options], [pytest], or [tool:pytest] scopes block default discovery;
• non-pytest pyproject.toml content is hashed and can remain conformant;
• safe shell prelude handling;
• direct, wrapped, and quoted eval termination;
• heredoc payload rejection;
• directory-changing command rejection;
• exact, non-path-qualified shell-template handling;
• self-referential and cyclic needs rejection;
• non-empty node, edge, and route collections;
• authority.mode == audit_only and all consequential grants false.

Refs #230
Follow-up: #232

• CI discovery now requires at least one literal, unfiltered automatic source-change event; paths, paths-ignore, and event-type-only trigger bypasses are frozen negative regressions.
• Default pytest discovery now audits root modules and packages that shadow the running interpreter's full standard-library namespace as well as pytest dependencies; hostile typing.py and json/__init__.py fixtures are pinned.

[coderabbitai]

📝 Walkthrough

Walkthrough

The PR adds Lotus Family Conformance v0.1: a manifest-driven, audit-only repository auditor with PASS/DRIFT/UNKNOWN outcomes, snapshot evidence hashing, GitHub Actions workflow discovery, causal route validation, regression tests, documentation, and automated CI execution.

Changes

Lotus Family conformance

Layer / File(s)	Summary
Audit contract and repository manifest standards/lotus-family/{README.md,ACCEPTANCE.md,DECISION-CODES.md,IMPLEMENTATION.md,NOTICE.md,ROADMAP.md,VERSION,.gitignore}, standards/lotus-family/manifest/*, standards/lotus-family/causality/*	Defines v0.1 outcomes, authority restrictions, evidence requirements, repository checks, CI discovery rules, and causal graph and route artifacts for Pythia, CML, and LS.
Snapshot audit engine standards/lotus-family/conformance/lotus_family_{schema,runtime,auditor,auditor_core}.py	Validates manifests, reads contained snapshot files, hashes evidence, evaluates checks, and exposes structured audit and CLI results.
GitHub Actions workflow discovery standards/lotus-family/conformance/lotus_family_workflow*.py	Parses workflow YAML and shell commands, extracts executable run steps, rejects indirect or unsafe test usage, and matches configured discovery strategies.
Causal graph and route validation standards/lotus-family/conformance/lotus_family_system_model.py, standards/lotus-family/conformance/test_causality_model.py	Validates graph topology and routes, computes centrality, renders traceability, executes route scenarios, and checks authority boundaries.
Regression coverage and CI execution standards/lotus-family/conformance/test_*.py, .github/workflows/lotus-family-conformance.yml	Adds fixture-based tests for audit outcomes, authority boundaries, hashes, workflow discovery, causal routes, and system-model behavior, and runs them through GitHub Actions.

Estimated code review effort: 4 (Complex) | ~60 minutes

Possibly related issues

• Issue 230 — Directly covers the Lotus Family conformance audit implemented by this PR, including manifests, repository adapters, outcomes, CI discovery, fixtures, and authority-boundary checks.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 17.06% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title clearly identifies the primary change: adding the Lotus Family auditor at version 0.1.
Description check	✅ Passed	The description covers all required sections, including changes, change types, rationale, exact-head validation, Lotus judgment criteria, compatibility and authority, and evidence. It provides the val…

Full details: Description check

Explanation

The description covers all required sections, including changes, change types, rationale, exact-head validation, Lotus judgment criteria, compatibility and authority, and evidence. It provides the validated commit SHA, validation command, test results, limitations, and pending hosted checks.

• Fix all pre-merge checks with AI

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch agent/lotus-family-auditor-v0-1

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[safal207]

@codex review

Please review exact head 22dee84fb1639fa461d182211a791471a74fff88, focusing on fail-closed UNKNOWN, CI-discovery false positives/negatives, evidence identity and hashing, manifest drift resistance, and the audit-only no-authority boundary.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 22dee84fb1

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".

[safal207]

Addressed all three Codex findings on exact head 33cfcaef44e42190d7d1f6af9140b95357945075:

• empty or missing manifest requirement lists now fail closed as UNKNOWN / MANIFEST_INVALID;
• CML CI discovery now requires a full pytest default-discovery invocation and rejects unrelated subsets, explicit contract-test ignores, and collection-only forms;
• manifest, workflow, test, and evidence paths are validated as repository-relative and resolved paths must remain inside the materialized snapshot.

Regression coverage increased from 7 to 12 tests, including each reported failure mode. The previous review and CI evidence for 22dee84fb1639fa461d182211a791471a74fff88 are superseded.

@codex review

Please review exact head 33cfcaef44e42190d7d1f6af9140b95357945075, focusing on the three fixes, fail-closed manifest validation, pytest discovery classification, path/symlink confinement, and the audit-only authority boundary.

[qodo-code-review]

ⓘ Qodo reviews are paused because your trial has ended. Ask your workspace admin to add credits to resume reviews. Manage billing

[safal207]

@coderabbitai review

Please review exact head 33cfcaef44e42190d7d1f6af9140b95357945075, focusing on fail-closed manifest validation, repository-snapshot path confinement, CI-discovery false positives/negatives, evidence hashing, and the audit-only no-authority boundary.

[coderabbitai]

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
set -euo pipefail

printf '%s\n' '== HEAD =='
git rev-parse HEAD
git status --short
printf '%s\n' '== Commit =='
git show -s --format='%H%n%P%n%s' HEAD

printf '%s\n' '== Changed files against first parent =='
git diff --name-status HEAD^ HEAD

printf '%s\n' '== Lotus-family file inventory =='
fd . standards/lotus-family -t f 2>/dev/null | sort || true

printf '%s\n' '== Python outlines =='
fd -e py . standards/lotus-family -t f -x ast-grep outline {} --items all --view expanded 2>/dev/null || true

printf '%s\n' '== Workflow outlines =='
fd -e yml -e yaml . .github -t f -x sh -c 'echo "--- $1"; sed -n "1,260p" "$1"' sh {} 2>/dev/null || true

printf '%s\n' '== Relevant source/config/test contents =='
fd -e py -e json -e yaml -e yml . standards/lotus-family -t f -x sh -c 'echo "--- $1"; cat -n "$1"' sh {} 2>/dev/null || true

Length of output: 63105

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

printf '%s\n' '== Recent commits affecting Lotus Family =='
git log --oneline --decorate -n 20 -- standards/lotus-family .github/workflows/lotus-family-conformance.yml

printf '%s\n' '== Candidate PR base refs and merge bases =='
git branch -a --contains HEAD || true
for ref in origin/main origin/master main master; do
  if git rev-parse --verify --quiet "$ref" >/dev/null; then
    printf '%s: ' "$ref"
    git merge-base "$ref" HEAD
  fi
done

printf '%s\n' '== Complete Lotus PR range (from likely merge base) =='
if git rev-parse --verify --quiet origin/main >/dev/null; then
  base="$(git merge-base origin/main HEAD)"
  printf 'base=%s\n' "$base"
  git diff --name-status "$base"...HEAD
fi

printf '%s\n' '== Lotus Family documentation =='
fd -e md -e txt -e json . standards/lotus-family -t f -x sh -c 'echo "--- $1"; cat -n "$1"' sh {}

printf '%s\n' '== Static CI classifier bypass analysis (standalone; does not import repository code) =='
python3 - <<'PY'
import shlex

PYTEST_START = ("python", "-m", "pytest")
test_path = "tests/test_lotus_docs_contract.py"
commands = [
    "echo python -m pytest",
    "python -m pytest --ignore=tests",
    "python -m pytest --deselect=tests/test_lotus_docs_contract.py::test_boundary",
    'python -m pytest -k "not lotus"',
    "python -m pytest --collect-only",
    "python -m pytest tests/test_other.py",
]
for command in commands:
    tokens = shlex.split(command)
    start = next(
        (i + 3 for i in range(len(tokens) - 2)
         if tuple(tokens[i:i + 3]) == PYTEST_START),
        None,
    )
    accepted = False
    if start is not None:
        args = tokens[start:]
        accepted = (
            not any(arg in {"--collect-only", "--co", "--setup-only"} for arg in args)
            and not any(test_path in arg for arg in args)
            and all(arg.startswith("-") for arg in args)
        )
    print(f"{accepted=!s:5}  {command}")
PY

Length of output: 15641

---

@safal207 Reviewed exact head 33cfcaef44e42190d7d1f6af9140b95357945075.

I found two correctness issues that can produce PASS without verifying the intended invariant:

1. Exact snapshot identity is not verified — audit_repository() only checks that commit_sha matches the 40-character lowercase-hex format (Lines 380–392), then audits whichever directory happens to be under snapshot_root (Lines 394–426). The regression fixture demonstrates this: arbitrary temporary content passes while reporting "a" * 40 as its commit (test Lines 69–82).
   This conflicts with the documented PASS meaning (“verified … exact commit SHA”) and UNKNOWN requirement when identity cannot be verified. Require a trusted, local provenance/attestation record that binds repository, ref, SHA, and snapshot content (or a verified local Git checkout), and return UNKNOWN when it is absent or mismatched.

2. CML pytest CI discovery remains bypassable — _shell_commands() scans raw workflow text rather than executable YAML run steps (Lines 262–279), and _is_pytest_default_discovery() permits any option-only invocation (Lines 282–305). Consequently, both of these classify as a passing discovery rule:

   # python -m pytest
   - run: echo python -m pytest
   - run: python -m pytest --ignore=tests

   The latter excludes tests/test_lotus_docs_contract.py, while the former does not run pytest at all. Reject shell prefixes/non-executable text, and reject collection-limiting options such as --ignore (including parent directories), --deselect, --confcutdir, and equivalent configuration-driven exclusions. Add regressions for these forms.

The manifest validation, contains_all fail-closed behavior, resolved-path confinement, evidence hashing of checked files, and stated audit-only authority boundary are otherwise aligned with the requested focus.

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 33cfcaef44

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".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8bce1ddf0e

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".

[safal207]

Exact-head follow-up: 530a808300527d6f10217321b4cbf964fcf002aa (tree 97a8f20b3aed53c6b7fc816b26506bc5dd92b13c). The three fresh discovery-proof findings are fixed with regressions; local Lotus conformance replay is 91/91 green. Hosted checks are intentionally treated as pending for this new head. @codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 530a808300

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".

[safal207]

Exact-head follow-up published at 0862cf6bf9ed3de3564c955459a71af4bb45598b (tree c57ed74fa26702c28087941f5b77603a4a280f40). Local conformance replay: 94/94 passed. The three fresh review findings are covered by regressions and their threads are resolved. Hosted CI and review must be evaluated only for this exact head.

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 0862cf6bf9

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".

[safal207]

Exact-head follow-up is published at 3bb3fedf9b60a3645603af4c2a9deb5f40aefc8e (tree bf52074b3c1282fb0ddbc35b3450648eb63f0349).

The fresh findings are addressed with bounded prerequisite trust: arbitrary setup is fail-closed; action prerequisites require explicit manifest trust plus immutable full SHAs; the configured Pythia CI remains discoverable; and the ordinary literal Mix wrapper plus literal required files are covered and hashed.

Local conformance: 98/98 passed; JSON, compile, and diff checks passed. Mix is unavailable locally, so hosted CI remains authoritative. Earlier-head evidence is stale and merge remains blocked pending exact-head hosted checks and fresh review.

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 3bb3fedf9b

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".

[safal207]

Exact-head update: c9d7bbb0f41c206cbeb9aa4c451c7d679b70bbce (tree a03168b28ab34d29d0a9d5070bc99d60f0684622). The Pythia Lotus gate now runs directly under Elixir before any Mix bootstrap; executable Mix project values, executable test helpers, and all required Mix files fail closed in generic Mix-default audits. Local replay: 101/101 conformance tests, Ruff, compileall, JSON, and diff checks pass. Elixir/Mix execution remains hosted-authoritative because those binaries are unavailable locally. Earlier-head results are stale. @codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c9d7bbb0f4

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".

[safal207]

@codex review

Exact head: c7a76da170d56e8f01cbcb4298c569a3c0fdf9d5.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c7a76da170

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".

[safal207]

@codex review

Please review exact head b4eded4d36dd06fcda278bbf934e6a2dadb55139 (tree 96219d9820c31090959fbfa957da859c92bfab92). Focus on the new all-language executed-source SHA binding and fail-closed pytest bytecode/native/startup shadow detection. Local canonical conformance replay: 109/109.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: b4eded4d36

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".

[safal207]

@codex review

Please review exact head 8031324e1be988f6f8ae79b3628ac90396d46b83 (tree 523a0b1cc626d7d5c9a5884ac0f7f259c8b4fefe). Focus on the closed pytest/runtime-dependency import-shadow families and their source/bytecode/native coverage. Local canonical conformance replay: 109/109.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8031324e1b

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".

[safal207]

@codex review

Please review exact head 0b2195239baa6d1dc5467bbba783d1d19d9e9643; all earlier-head evidence is stale.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 0b2195239b

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".

[safal207]

@codex review

@coderabbitai review

Fresh exact-head review requested for 30d9f018e88a1a947a88406228cc7e91b9111218.

[coderabbitai]

Tip

For best results, initiate chat on the files or code changes.

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
set -euo pipefail

expected_head='30d9f018e88a1a947a88406228cc7e91b9111218'

printf '%s\n' '== Exact revision =='
actual_head="$(git rev-parse HEAD)"
printf 'HEAD=%s\n' "$actual_head"
printf 'TREE=%s\n' "$(git rev-parse HEAD^{tree})"
git show -s --format='subject=%s%nparents=%P' HEAD
if [ "$actual_head" != "$expected_head" ]; then
  printf 'ERROR: expected %s but checked out %s\n' "$expected_head" "$actual_head" >&2
  exit 4
fi

printf '%s\n' '== Changed files at exact head =='
git diff --name-status HEAD^ HEAD

printf '%s\n' '== Lotus conformance Python inventory =='
fd -t f -e py . standards/lotus-family/conformance | sort

printf '%s\n' '== Public-path and test outlines =='
fd -t f -e py . standards/lotus-family/conformance \
  -x sh -c '
    case "$1" in
      *workflow*.py|*runtime*.py|*schema*.py|test_*.py)
        echo "--- $1"
        ast-grep outline "$1" --items all --view expanded
        ;;
    esac
  ' sh {} 2>/dev/null

printf '%s\n' '== Current public imports and call sites =='
rg -n -C 3 \
  'from lotus_family_workflow|import lotus_family_workflow|\bci_discovery\s*\(|\bgithub_run_scripts\s*\(|\bshell_commands\s*\(|\baudit_repository\s*\(|identity_assurance|grants_(ownership|approval|execution|delivery|deployment|merge)' \
  standards/lotus-family/conformance -g '*.py'

printf '%s\n' '== Focused policy/runtime/schema content =='
fd -t f -e py . standards/lotus-family/conformance \
  -x sh -c '
    case "$1" in
      *lotus_family_workflow_policy*.py|*lotus_family_workflow.py|*lotus_family_runtime*.py|*lotus_family_schema.py)
        echo "--- $1"
        cat -n "$1"
        ;;
    esac
  ' sh {}

printf '%s\n' '== Boundary regression references =='
rg -n -C 4 \
  'eval|heredoc|terminator|command -|builtin -|continue-on-error|needs|PYTEST_|pytest|conftest|pytest\.ini|pyproject|provenance|caller_claim_only|authority|grants_' \
  standards/lotus-family/conformance -g 'test_*.py'

printf '%s\n' '== Manifest and workflow content =='
fd -t f -e json . standards/lotus-family/manifest -x sh -c 'echo "--- $1"; cat -n "$1"' sh {}
fd -t f -e yml -e yaml . .github/workflows -x sh -c 'echo "--- $1"; cat -n "$1"' sh {}

Length of output: 160878

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

printf '%s\n' '== Exact revision =='
git rev-parse HEAD
git rev-parse HEAD^{tree}

printf '%s\n' '== Public workflow import chain =='
for file in \
  standards/lotus-family/conformance/lotus_family_workflow.py \
  standards/lotus-family/conformance/lotus_family_workflow_policy.py \
  standards/lotus-family/conformance/lotus_family_runtime.py \
  standards/lotus-family/conformance/lotus_family_schema.py
do
  if [ -f "$file" ]; then
    echo "--- $file"
    cat -n "$file"
  fi
done

printf '%s\n' '== Versioned workflow implementations =='
fd -t f -e py 'lotus_family_workflow_policy*.py' \
  standards/lotus-family/conformance \
  -x sh -c 'echo "--- $1"; rg -n -C 5 "_SHELL_WRAPPERS|_TERMINATORS|def _unsafe_state_change|def _analyze_script|def _ci_discovery_one|def ci_discovery" "$1"' sh {}

printf '%s\n' '== Wrapper-option coverage =='
rg -n -C 5 \
  'command\s+-p|command\s+--|builtin\s+-|builtin\s+--|command exit|builtin exit|wrapped.*terminator|wrapper.*option|_unsafe_state_change' \
  standards/lotus-family/conformance -g '*.py'

printf '%s\n' '== Provenance, configuration, evidence, and authority paths =='
rg -n -C 5 \
  'def (validate_manifest|read_file|result|manifest_invalid|audit_repository)|caller_claim_only|identity_assurance|provenance|pytest_configuration|mix_configuration|grants_(ownership|approval|execution|delivery|deployment|merge)' \
  standards/lotus-family/conformance -g '*.py'

printf '%s\n' '== Exact-head patch =='
git diff --find-renames --find-copies HEAD^ HEAD -- standards/lotus-family/conformance

Length of output: 104014

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

printf '%s\n' '== Exact revision =='
git rev-parse HEAD

printf '%s\n' '== Public v4 policy =='
cat -n standards/lotus-family/conformance/lotus_family_workflow_policy_v4.py

printf '%s\n' '== v4 imports and shared analyzer call paths =='
rg -n -C 6 \
  'import .*workflow_policy|from .*workflow_policy|_analyze_script|_unsafe_state_change|_github_run_step_groups|shell_commands|_ci_discovery_one|def ci_discovery' \
  standards/lotus-family/conformance/lotus_family_workflow_policy_v4.py \
  standards/lotus-family/conformance/lotus_family_workflow_policy.py \
  standards/lotus-family/conformance/lotus_family_workflow_policy_v2.py \
  standards/lotus-family/conformance/lotus_family_workflow_policy_v3.py

printf '%s\n' '== Public v4 wrapper and terminator regression coverage =='
rg -n -C 6 \
  'command\s+-p|command\s+--|builtin\s+-|builtin\s+--|command exit|builtin exit|wrapped.*terminator|wrapper.*option|eval.*exit|exit 0' \
  standards/lotus-family/conformance -g 'test_*.py'

printf '%s\n' '== v4 workflow event and container regression coverage =='
rg -n -C 5 \
  'workflow_dispatch|pull_request|merge_group|automatic.*trigger|job_container|container:' \
  standards/lotus-family/conformance -g 'test_*.py'

Length of output: 61532

---

@safal207 Reviewed exact head 30d9f018e88a1a947a88406228cc7e91b9111218.

I found no actionable findings in the inspected public paths.

The public API exports lotus_family_workflow_policy_v4. The new policy rejects manual-only workflows and job containers. The v4 policy also fails closed when an earlier shell command is not a trusted prerequisite.

The shared analyzer does not skip command or builtin options before it checks wrapped terminators. However, this does not create a public v4 discovery bypass. A command such as command -p exit 0 is not an allowed prerequisite. The v4 policy stops discovery before a later test command.

The auditor retains identity_assurance.mode = "caller_claim_only". It does not claim remote repository, commit reachability, or working-tree verification. The audit result retains audit_only authority. Ownership, approval, execution, delivery, and merge grants remain false.

I performed static review only. I did not run repository code or tests.

_{You are interacting with an AI system.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 30d9f018e8

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".

[safal207]

Fresh exact-head review requested for 95d0c15a6a52ac84224ff830ba679b624133bfae (bfa56e436f8af2fae2577c29e8a984fddedffd59). Please verify unfiltered source-trigger coverage and full-stdlib pytest import-shadow handling. Do not merge until all checks and fresh review are clean on this exact head.