fix: close deterministic agent-artifact intake normalization lifecycle by Stahl-G · Pull Request #484 · Stahl-G/briefloop

Stahl-G · 2026-07-10T14:44:02Z

Merge Unit

MU-3: deterministic agent-artifact intake normalization lifecycle.

Final head: 975f9390846bc642fa12e83fdd44ce60dc28753e

Base after the path-authority predecessor (#487):
3834af60e24dde0fba43aea8f50e91d0ce0a0092

Complete PR diff: 29 files, +5898/-373.

Opening Contract

Invariant:
  candidate_claims, screened_candidates, and claim_drafts are raw agent
  proposals. One deterministic evaluator owns bounded mechanical
  normalization, canonical normalized bytes, digests, strict validation, and
  typed findings. The artifact registry persists the current-run intake
  projection. Deterministic consumers use that projection/bundle and never
  treat normalized validity as permission to clear repair/supersede staleness.

Producer:
  multi_agent_brief.contracts.agent_artifact_intake

Authority records:
  artifact_registry.json artifacts[artifact_id].intake_projection
    schema: briefloop.intake_projection.v1
  runtime_manifest.json claim_ledger_freeze
    new writes: mabw.claim_ledger_freeze.v2
    legacy reads: explicit v1 lineage validation only

Writers/recomputers:
  intake evaluator; artifact registry recomputer; Claim Ledger freeze and
  metadata-enrichment transactions.

Consumers:
  registry validation; Scout/Screener stage and topology completion; screened
  universe validation; Claim Ledger freeze/repeat/completion; coverage gates;
  materiality projection; status diagnostics.

Consumers may:
  read raw proposals; consume normalized payloads and persisted diagnostics;
  reject invalid, stale, mismatched, cross-run, or malformed authority.

Consumers may not:
  rewrite raw proposals; invent evidence identity; map ambiguous semantics;
  maintain local alias/hash/universe rules; infer current validity from file
  presence; clear stale_after_repair or stale_after_supersede.

Run/transaction identity:
  registry run_id equals manifest run_id; projections bind artifact_id, raw
  SHA, normalized SHA, transform version, counts, and typed entries; freeze v2
  binds current run, transaction, source projection, paths, event, and ledger.

Legacy behavior:
  no old-workspace migration or silent upgrade. Unknown schema/transform and
  missing authority fail closed. A supported v1 freeze must prove current-run
  event lineage; post-freeze enrichment must prove immutable old-to-new hashes.

Path boundary:
  canonical path production and generic gate-consumer migration are inherited
  from merged predecessor #487. This PR adds only the agent-intake selector and
  intake/freeze consumer bindings; it does not create a second path authority.

Non-goals / deferred:
  registry sibling-inference debt outside the migrated gate-impacting paths;
  status/Panel wording redesign; old-workspace migration; hardlink/casefold
  platform policy beyond #487; renderer/finalize-audit path migration; MU-4.

State x Path Matrix

ID	State / path	Deterministic outcome	Representative test
INTAKE-PARSE-01	canonical candidate payload	valid; no normalization; both digests	`test_candidate_intake_canonical`
INTAKE-PARSE-02	malformed JSON	typed fatal parse finding	`test_candidate_intake_malformed_json`
INTAKE-ALIAS-01	canonical/alias conflict	exact-path fatal finding	`test_candidate_intake_alias_conflict`
INTAKE-ID-01	missing/duplicate candidate ID	fail closed	`test_candidate_intake_requires_candidate_id`; `test_candidate_intake_duplicate_ids`
INTAKE-ID-02	malformed supplied source identity	shared validators reject	`test_legacy_candidate_supplied_identity_fields_are_validated`
INTAKE-SCREEN-01	selected/excluded mechanical aliases	one normalized screening shape	`test_screened_bucket_aliases`
INTAKE-SCREEN-02	prose reason without reason_code	no consumer-local inference	`test_screened_reason_not_reason_code`
INTAKE-UNIVERSE-01	unknown/duplicate/missing/count mismatch	registry, coverage, materiality share one invalid verdict	`test_workspace_intake_bundle_owns_screened_universe_verdict`
INTAKE-UNIVERSE-02	missing candidate universe	screened intake fails closed	`test_workspace_intake_bundle_rejects_missing_candidate_universe`
INTAKE-UNIVERSE-03	registry-invalid screening reaches quality gate	no report and no pass event	`test_coverage_gate_rejects_screened_candidate_universe_mismatch`
INTAKE-PATH-01	custom candidate/screened paths; poisoned defaults	quality/materiality consume contract bindings	`test_quality_and_materiality_bind_contract_resolved_intake_paths`
INTAKE-DRAFT-01	bounded draft aliases	canonical drafts; raw unchanged	`test_claim_draft_mechanical_normalization`
INTAKE-DRAFT-02	unknown claim type or nested claim_id	fatal before freeze	`test_claim_draft_unknown_claim_type`; `test_claim_draft_nested_claim_id`
INTAKE-REGISTRY-01	malformed/impossible projection	persisted authority rejected	`test_intake_projection_internal_binding_fails_closed`; `test_projection_rejects_impossible_status_digest_finding_combination`
INTAKE-REGISTRY-02	transplanted artifact projection/status	artifact identity/status binding rejected	`test_registry_intake_projection_binds_artifact_identity_and_status`
INTAKE-REGISTRY-03	cross-run or unknown version	fail closed	`test_registry_intake_projection_rejects_cross_run_context`; `test_registry_intake_projection_rejects_unknown_version`
INTAKE-STALE-01	intake-valid plus repair/supersede stale	stale remains non-consumable	`test_screened_intake_consumption_rejects_stale_registry_dependencies`
INTAKE-STALE-02	intake-invalid plus stale baseline	invalid > stale > valid	`test_invalid_intake_precedes_recovery_stale_overlay`
INTAKE-FREEZE-01	real freeze from normalized drafts	v2 binds raw/normalized/registry/event identities	`test_freeze_binds_intake_projection`
INTAKE-FREEZE-02	registry/evaluator mismatch	reject before writes	`test_freeze_rejects_projection_mismatch`
INTAKE-FREEZE-03	repeat raw/policy drift	reject even if normalized bytes appear equivalent	`test_repeat_freeze_rejects_raw_drift`; `test_repeat_freeze_rejects_policy_drift`
INTAKE-FREEZE-04	partial event/state failure	atomic rollback	`test_freeze_intake_binding_rolls_back_atomically`
INTAKE-FREEZE-05	stale claim_drafts	repeat freeze and stage completion blocked	`test_stale_claim_drafts_blocks_repeat_freeze_and_stage_completion`
INTAKE-LEGACY-01	v1 freeze transplanted cross-run	reject without matching current-run event	`test_legacy_freeze_rejects_cross_run_transplant`
INTAKE-LEGACY-02	duplicate/wrong freeze event	require exactly one matching stage/artifact event	`test_freeze_binding_rejects_duplicate_transaction_events`; `test_freeze_binding_rejects_wrong_event_identity`
INTAKE-LEGACY-03	v1 freeze then deterministic enrichment	immutable event anchors old hash; lineage binds new hash	`test_legacy_freeze_remains_valid_after_metadata_enrichment`
INTAKE-LEGACY-04	corrupted/missing enrichment lineage	fail closed	`test_legacy_freeze_rejects_invalid_metadata_enrichment_lineage`
INTAKE-PATH-02	custom drafts/Ledger through freeze and enrichment	registry/manifest/event share contract identity; defaults untouched	`test_metadata_enrichment_binds_claim_paths_from_artifact_contracts`
INTAKE-PATH-03	finalize gate metadata uses custom Ledger	completion quality-binding uses the same contract Ledger	`test_metadata_enrichment_binds_claim_paths_from_artifact_contracts`
INTAKE-STATUS-01	valid and corrupted projections	compact diagnostics; corruption unavailable	`test_status_formats_intake_projection`; `test_status_rejects_invalid_intake_projection`
INTAKE-E2E-01	public-safe recoverable/fatal fixtures	recoverable progresses; identity fatal is stable	`test_public_safe_recoverable_intake_fixture`; `test_public_safe_fatal_intake_fixture`

Consumer Sweep

CLI/runtime stage completion: migrated for agent artifacts.
Registry/state recompute: one dependency-ordered intake bundle and persisted projection.
Claim Ledger freeze/repeat/enrichment/completion: current-run and event-bound.
Quality coverage/materiality/status: same bundle plus registry stale eligibility.
Generated Scout assets: source role contract updated first; generated mirrors in sync.
WorkBuddy/Hermes adapters remain non-authoritative raw-proposal readers.
Generic path producer and CSM/atomic/auditor path consumers come from fix: bind gate consumers to artifact path authority #487.

Validation

Final-head local results:

focused intake/materiality/quality/status: 151 passed
runtime state lifecycle: 414 passed
full repository suite: 2986 passed
Ruff on all changed Python files: passed
generated agent configs: passed
version consistency: passed
release consistency (--no-tag): passed
capabilities (current src): passed
product baseline: passed
public safety: passed
git diff --check: passed
README.md and README_en.md: inspected; unchanged

Authority Declaration

New independent authority file: no
Authority schema/semantics changed: yes
Authority lifecycle/consumer binding changed: yes
Raw proposal mutation: no
Old workspace migration: no

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 20d62ad562

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 · 2026-07-10T17:15:02Z

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: bf72ab1ca7

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: aa06ddc60e

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: 4744c31376

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".

cursor

Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 325588e. Configure here.}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e5a7a7c37a

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: cf980fb59b

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".

cursor · 2026-07-11T01:34:01Z

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_046cf662-4303-4db9-9d90-f1e7626ff4ff)

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 59bef1238e

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".

…inal

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 06a6d8c81b

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: 4fd8188285

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".

# Conflicts: # src/multi_agent_brief/orchestrator/runtime_state/artifact_paths.py # src/multi_agent_brief/orchestrator/runtime_state/artifact_registry.py # src/multi_agent_brief/orchestrator/runtime_state/claim_support_matrix.py # src/multi_agent_brief/orchestrator/runtime_state/completion_gates.py # src/multi_agent_brief/orchestrator/runtime_state/contracts_loader.py # src/multi_agent_brief/quality_gates/state.py # tests/test_runtime_state.py

Stahl-G · 2026-07-11T11:46:23Z

@codex review

Stahl-G · 2026-07-11T11:46:40Z

Northstar product review

Decision: DO NOW / accept the bounded MU-3 slice, subject to final CI and review.

User and job: a BriefLoop operator needs agent-authored workflow artifacts to cross one deterministic intake boundary without registry, gate, freeze, and status surfaces giving different answers.
Evidence: repeated reproducible control-plane defects and adversarial lifecycle cases. This is engineering/safety evidence; it is not evidence that report quality or user value improved.
Product invariant: agents propose; deterministic Python validates, records, freezes, and gates. Normalized validity cannot erase repair/supersede staleness.
MVP: one evaluator, one persisted current-run projection, bound freeze lineage, and the in-scope completion/coverage/materiality/status consumers.
Non-goals retained: no old-workspace migration, second authority, new user-facing concept, broader filesystem governance, renderer/finalize-audit migration, or public capability upgrade.
Public-claim boundary: this PR supports a stronger deterministic control invariant only. It does not justify claims of better factual accuracy, semantic quality, or Pilot readiness.

The predecessor path-authority work is now in #487/main; the final #484 diff removes the duplicate path-authority package and keeps only the thin intake binding needed by MU-3.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 975f939084

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 · 2026-07-11T11:51:36Z

+    artifact_paths = agent_artifact_paths_from_contracts(
+        ws,
+        artifact_records if isinstance(artifact_records, Mapping) else {},
+    )


Handle invalid registry paths in status projections

When briefloop status reads a corrupt or tampered artifact_registry.json whose agent artifact path is unsafe (for example ../evil), this call raises RuntimeStateError from the artifact path resolver and aborts the whole read-only status summary. build_workspace_status is documented to only read existing files and report corrupt surfaces as stale/unknown, so the materiality projection should catch this and return an invalid_screened_candidates/unavailable projection instead of letting a bad registry path crash status.

Useful? React with 👍 / 👎.

Merge Governor added 2 commits July 10, 2026 21:43

fix: add deterministic agent artifact intake evaluator

d81be66

fix: route artifact consumers through intake projection

b292722

Stahl-G mentioned this pull request Jul 10, 2026

fix: scope recovery cycles and gate stage completion #485

Merged

Merge Governor added 4 commits July 10, 2026 22:13

fix: bind claim ledger freeze to intake projection

6273b9d

fix: expose persisted intake diagnostics

23322bd

test: fail closed on malformed intake projections

f4947fa

refactor: keep intake status projection boolean

20d62ad

Stahl-G marked this pull request as ready for review July 10, 2026 15:59

chatgpt-codex-connector Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/contracts/agent_artifact_intake.py Outdated

fix: close agent artifact intake authority gaps

7ee7a32

cursor Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/orchestrator/runtime_state/completion_gates.py Outdated

fix: handle mixed completion artifact gates

bf72ab1

cursor Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/orchestrator/runtime_state/artifact_registry.py

fix: preserve invalid intake over stale overlay

aa06ddc

chatgpt-codex-connector Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/orchestrator/runtime_state/completion_gates.py Outdated

chatgpt-codex-connector Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/contracts/agent_artifact_intake.py

fix: bind intake lifecycle to contract paths

4744c31

cursor Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/product/materiality_selection.py Outdated

chatgpt-codex-connector Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/quality_gates/state.py Outdated

fix: propagate intake contract paths to consumers

325588e

cursor Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/orchestrator/runtime_state/claim_ledger_freeze.py

fix: bind freeze to contract artifact paths

e5a7a7c

chatgpt-codex-connector Bot reviewed Jul 10, 2026

View reviewed changes

Comment thread src/multi_agent_brief/orchestrator/runtime_state/claim_ledger_freeze.py Outdated

fix: bind legacy freeze to enrichment lineage

cf980fb

chatgpt-codex-connector Bot reviewed Jul 11, 2026

View reviewed changes

Comment thread src/multi_agent_brief/orchestrator/runtime_state/claim_ledger_freeze.py Outdated

fix: close artifact path and intake identity authority

59bef12

chatgpt-codex-connector Bot reviewed Jul 11, 2026

View reviewed changes

Comment thread src/multi_agent_brief/orchestrator/runtime_state/artifact_registry.py Outdated

Merge remote-tracking branch 'origin/main' into codex/pr484-minimal-f…

664112d

…inal

fix: close bounded MU-3 intake consumption paths

06a6d8c

chatgpt-codex-connector Bot reviewed Jul 11, 2026

View reviewed changes

Comment thread src/multi_agent_brief/orchestrator/runtime_state/claim_ledger_freeze.py Outdated

fix: bind artifact path consumers to contracts

4fd8188

chatgpt-codex-connector Bot reviewed Jul 11, 2026

View reviewed changes

Comment thread src/multi_agent_brief/quality_gates/state.py

Stahl-G mentioned this pull request Jul 11, 2026

fix: bind gate consumers to artifact path authority #487

Merged

Stahl-G merged commit 3042927 into main Jul 11, 2026
8 checks passed

chatgpt-codex-connector Bot reviewed Jul 11, 2026

View reviewed changes

Conversation

Stahl-G commented Jul 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Merge Unit

Opening Contract

State x Path Matrix

Consumer Sweep

Validation

Authority Declaration

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot commented Jul 10, 2026

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

cursor Bot commented Jul 11, 2026

Bugbot couldn't run - usage limit reached

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Stahl-G commented Jul 11, 2026

Uh oh!

Stahl-G commented Jul 11, 2026

Northstar product review

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Jul 11, 2026

Choose a reason for hiding this comment

Uh oh!

Stahl-G commented Jul 10, 2026 •

edited

Loading