Skip to content

Integration: all groups combined; every rule Implemented or N/A - #510

Merged
sam-aydlette merged 72 commits into
feat/frr-recordsfrom
feat/integration
Nov 6, 2026
Merged

sam-aydlette merged 72 commits into
feat/frr-recordsfrom
feat/integration

Conversation

@sam-aydlette

@sam-aydlette sam-aydlette commented Nov 5, 2026 •

Copy link
Copy Markdown
Owner

SCN-Type: adaptive
SCN-Rationale: adds two IAM roles for unattended publishing, runtime-emitter grants, CloudFront access logging and a DAST scanner user; no new external service, no federal data, no class change.
SCN-Record: SCN-2026-007

Changed

Merges (plain merges into a branch cut from feat/frr-records): feat/trust-scope-assessor (#509, which carries #503's schema-shaped package), then feat/notifications (#508), then feat/vdr-followups (#505).

How the conflicts were resolved:

  • data/cds/offering.json: kept the schema-form values and added the FedRAMP Security Inbox block. The personal email address from the notifications side was not carried over.
  • docs/boundary/boundary-classification.json: kept flows L to Y4 and renumbered the incident-report flow to Z. Added the incident role and policies to the MAS fixture.
  • infrastructure/schemas/freshness-config.json: combined both sides' entries (inbox simulation, SCN JSON, SCN Atom feed, SCN evaluations). The certification overview stays retired.
  • scripts/build-ocr.py:
    • combined the FedRAMP-schema OCR with the incident attestation, which is read tolerantly;
    • both branches declared --incidents, so it is now declared once, with a test (below).
  • scripts/build-cpo.py:
    • the schema-shaped CPO now carries the SCN notification mechanism and the FedRAMP Security Inbox, in providerExtensions;
    • the SCN and incident files are listed in additionalRepositories;
    • the SCN and inbox tests now read the schema-shaped CPO.
  • data/frr-records.json and data/decisions.json: merged record by record, each record taken from the branch that owns it.
  • policies/triage.rego: kept the grouped decisions and added the incident triggers.
  • scripts/reconcile.py: both branches added an invariant (m). Incident findings stay (m), and runtime V&V coverage becomes (n); the three records that cite it were updated.
  • deploy-with-opa.yml, evidence-nightly.yml: kept every step from both sides, including both preflights and both post-deploy jobs, combined the redaction lists, and kept the publish lists equal to the gate. The SCN build and the FRC publish now run before the SDR build.
  • data/non-machine-resources.json: added the FedRAMP Security Inbox runbook (DEC-190).

Records closed:

  • CDS-CSO-FID:
    • scripts/_offering.stamp_fedramp_id stamps every freshness-governed JSON artifact: in providerExtensions for FedRAMP-schema reports, at the top level for site files;
    • check-fedramp-id.py checks the VDR family, the KSI signal, the SCN and incident files, the inbox simulation, MAS, SCG, settings, the access summary, the boundary map, the reconcile report and conformance;
    • four files are exempt, each with a stated reason;
    • a test fails when a governed artifact is neither checked nor exempt;
    • the nightly and the incident publisher run the gate on what they republish;
    • the runtime Lambda copies the ID into the runtime signal.
  • FRC-CSX-VVK:
    • build-sdr.py counts dependency scanning for KSIs that have npm or PyPI components, only when the bound VDR is signed by the deploy or nightly identity, fresh, and every Grype feed passed;
    • it counts an automated check of the training attestation for KSI-CED-RAT (DEC-192, in force, decided under 12 months ago);
    • KSI-CED-RAT's single method is accepted in DEC-191, and the record says so plainly;
    • the deploy holds it:
      • after apply, the deploy stages its signed inventory under deploy-staging/<sha>/ and invokes the runtime emitter synchronously with {"stage": "<sha>"};
      • the emitter accepts only a 40-hex SHA, re-validates the deployed system against that inventory, and writes its signed signal beside it, using a new PutObject grant on that one key pattern in main.tf;
      • the deploy reads the signal back, deletes the staging objects, and checks that it re-validated this commit;
      • the VDR and SDR use that signal: build-sdr.py --runtime-commit counts the runtime method only for it, and --require-two-methods fails the deploy before anything is published;
      • the package maintainer applies the same requirement daily.
  • FRC-CSO-JSN: SCN notifications and incident reports go through write_validated. The record now cites the schema tests and the known-bad tests (a document that fails is not written).
  • process-vdr.unresolved-component:
    • aws_iam_group_policy, aws_iam_role_policy_attachment and aws_iam_group_policy_attachment are now inventory components. An attachment block is one component, with its policies listed by name only.
    • Address resolution tries the exact resource type before the prefix match, so aws_iam_group_policy_attachment.read_only resolves to the attachment, not to the group.
    • No suppression decision was added.

Operator decisions applied:

  • security.txt and the KSI signal's operator_contact publish the operator's LinkedIn profile instead of a personal email address, with a test.
  • The change is adaptive: SCN-2026-007 is in the register with its impact analysis (AI-drafted).

Found during integration:

  • Duplicate --incidents option: the merged build-ocr.py declared it twice, which would have stopped the deploy at the OCR step. A new test runs --help on every argparse script a workflow calls, with a known-bad fixture.

  • IAM user names: the vulnerability history backfill, the trust-center register and the VDR's suppression scopes published decision resources, including IAM user names (aws-iam-user::<name>). Published copies now carry aws-iam-user::NAME-WITHHELD, and the publication redaction gate fails on any remaining IAM user name, with a known-bad fixture.

  • GitHub run IDs are no longer published:

    • the nightly beacon's run_url is replaced by run: {workflow, commit};
    • the KSI signal drops provenance.builder.run_id;
    • the SLSA provenance omits runDetails.metadata.invocationId, which SLSA v1.0 does not require (BuildMetadata has no required field);
    • check-published-redaction.py fails on a run URL or a numeric run_id, with known-bad fixtures.

    No consumer read these values. The SLA-alarm runbook now finds the run by workflow and time.

  • CodeGuard: the SDR step passes the VDR signer identity through env instead of interpolating an expression into the script.

Effect

After deploy:

  • every published report carries the FedRAMP ID;
  • the SDR publishes per-KSI automatedMethods, packageComponents and fewerMethodsAccepted;
  • the bootstrap group policies and policy attachments appear in the inventory, and their checkov findings resolve to them;
  • no published file names an IAM user.

Rule records, from python3 scripts/_frr.py:

Implemented Partially Not Implemented Not Applicable
feat/frr-records (base) 6 93 42 13
this branch 141 0 0 13
"rules": 154,
"by_status": {"Implemented": 141, "Partially Implemented": 0, "Not Implemented": 0, "Not Applicable": 13},
"simulated": {"agency": 4, "assessor": 7, "fedramp": 2, "incident-recipient": 4, "mail": 7, "sam-gov": 1}

python3 scripts/_frr.py --complete exits 0: every applicable rule is Implemented or Not Applicable.

Verified

  • make check passes: 1834 tests, 1 skipped, plus ruff, OPA tests and coverage.
  • make check-complete exits 0.
  • These all pass: Regal lint of infrastructure/policy (no violations), check-policy-annotations.py, check-exceptions.py, check-wasm-safe.py.
  • storycheck, on a local build of the CPO, SDR, OCR, VDT, AVI and history from the live signal and VDR: FMT-1 to FMT-4 pass (10 pass, 0 fail).
  • Scan of the locally built published files (the six reports plus the SCN, incident, delivery, inbox-simulation and NMV outputs):
    • no twelve-digit numbers;
    • no ARNs with an account;
    • no run URLs;
    • no email addresses other than allowed domains;
    • no IAM user names after the fix above;
    • check-published-redaction.py passes over them.
  • Every new gate invariant has a unit test and a known-bad fixture: the FedRAMP ID coverage, the VVK method count and shortfall, the training-attestation age, unresolved bootstrap components, the IAM-user gate and duplicate argparse options.

Residual

  • VVK before the first deploy: an SDR built locally from the currently deployed (pre-merge) signal shows 11 KSIs short of two methods. The record's count (43 of 46 by plan and runtime, plus the two supply-chain KSIs by dependency scanning) comes from this branch's plan rules and runtime readers. If the staged run does not reach it, the deploy fails before publishing; no SDR contradicting the record is published.
  • Staged objects: the staged inventory and runtime signal sit briefly under deploy-staging/<sha>/ in the website bucket, so the CDN may serve them while they exist. They hold the same redacted content the deploy is about to publish, and are deleted on exit, whether the step succeeds or fails.
  • Run URIs in Sigstore bundles: every Sigstore bundle (*.bundle) and the provenance envelope (*.intoto.jsonl) carry the Fulcio certificate, whose Run Invocation URI extension names the GitHub run. That is part of keyless signing; removing it would mean dropping keyless signing. The publication gate does not scan Sigstore's output.
  • KSI signal identifiers: the KSI signal publishes the Cognito pool and client IDs (noted in Trust center, scope and simulated assessor (W8, W10) #509).
  • Bootstrap check findings: they now resolve to components, and they stay open findings on those components with their own deadlines.

Needs human

In order:

  1. Bootstrap applies (operator, one terraform apply of infrastructure/bootstrap from a tree synced to this branch):

    • package-maintenance.tf;
    • incident-reports.tf, including the deploy-environment trust;
    • cloudfront.tf (the /.well-known/* response-headers policy and behaviour);
    • trust-access-logs.tf;
    • deploy-config-reads.tf.
  2. Repository variables, from the bootstrap outputs:

    • AWS_PACKAGE_MAINTENANCE_ROLE_ARN from package_maintenance_role_arn;
    • AWS_INCIDENT_REPORTS_ROLE_ARN from incident_reports_role_arn.
  3. GitHub Environments, each with no reviewer and deployment branches restricted to exactly main:

    • package-maintenance;
    • incident-reports.
  4. DAST scanner (see docs/runbooks/dast-scanner-user.md): create the scanner user and the secrets SILK_DAST_USERNAME, SILK_DAST_PASSWORD, SILK_DAST_USER_POOL_ID and SILK_DAST_TOTP_SECRET.

  5. Labels: create ocr-feedback, quarterly-review-registration and package-maintenance.

  6. Account and repository checks:

    • confirm that account-level S3 Block Public Access is on;
    • confirm that no ruleset blocks the workflow's push to the availability-status branch.
  7. SCN lines: add them to the descriptions of FedRAMP 20x Class C: every applicable rule Implemented or Not Applicable #501 and of any open Dependabot PRs, or scn-tag fails.

  8. Review the AI-drafted text:

    • all 154 records, especially the sam items (application, UEI, assessor), the 13 Not Applicable reasons and the MAY rationales;
    • the issue-form wording, the simulated FedRAMP Certification Report, the privacy notice and the procedures;
    • the incident contact and recipient wording, the federal incident coordinator text and the exercise content.
  9. Confirm the AI-drafted decisions:

    • DEC-100..107, DEC-150..152, DEC-160, DEC-165..167, DEC-180..182 and DEC-190;
    • new here: DEC-191 (KSI-CED-RAT has one automated method, an accepted finding) and DEC-192 (the operator's training attestation);
    • whether DEC-190 covers the FedRAMP Security Inbox runbook added to the non-machine register;
    • DEC-007/009/010 detection dates taken from git history;
    • the gate-modes entries (vdr.coverage, vdr.unresolved-component, reconcile.m block);
    • the third-party fields in data/minimum-assessment-scope.json (ext::duo-mfa is inferred).
  10. Re-decide or amend:

    • DEC-010: its justification does not fit silk_reeling;
    • DEC-025: its planned_completion contradicts DEC-150;
    • the incident register policy's decided_by: Operator.
  11. Decide:

    • the unattended publishing paths (package maintenance, incident publisher) and widening the incident role's trust;
    • the freshness windows (SDR, CPO and OCR at 336 h; incident and delivery logs at 24 h);
    • SCN-2026-001's adaptive categorization and counting the emergency 5 business days from the merge;
    • NRI adoption;
    • whether to tell FedRAMP that the SCN feed exists (nothing is sent);
  12. Within the next weeks:

    • rotate the operator access key before about 2026-12-23;
    • remove the IAM user named with an outside person's email;
    • fix the stale CSP in infrastructure/main.tf;
    • correct the FIPS claim in authorization-boundary.html (an article; not edited here).
  13. Merge, then deploy:

    1. Merge into feat/frr-records, then main.
    2. Let CodeQL run on main.
    3. Dispatch a ZAP run.
    4. Wait for the first availability-probe run.
    5. Approve the deploy promptly: the header, logging and availability checks fail it until steps 1 to 4 are done, and invariant (n) fails nightlies until the new runtime Lambda is deployed.

    The main-stack apply happens in the deploy. It:

    • adds read-only grants to the runtime Lambda and raises its timeout to 180 s;
    • adds the 90-day log prefix rules;
    • adds the incident watchdog alarm;
    • rebuilds the watchdog from the new freshness config.

    Costs, all estimated: CloudFront log delivery under $0.05 a month, API scan traffic under $0.25 a month, one Cognito MAU (free tier). CodeQL, Actions and ZAP are free on a public repo.

Couldn't verify

  • Anything that needs AWS or GitHub Actions:
    • the bootstrap and main-stack applies;
    • the live header, logging and availability checks;
    • cosign verification of the VDR inside the SDR build (tested with an injected verifier);
    • the deploy's and maintainer's runs end to end;
    • the runtime emitter's fedRampId copy (checked statically).
  • The staged emitter run:
    • the deploy role invoked the emitter successfully in the last main deploy (status 200), so no bootstrap grant is needed;
    • the emitter's new PutObject grant is a main-stack change, applied by the deploy before the step runs;
    • the staged run itself was not exercised locally (it needs AWS).
  • Inventory changes against live state: whether the live bootstrap state lists exactly the group policies and attachments the fixture models. The fixture follows infrastructure/bootstrap/*.tf.
  • Builders that need AWS or a full build: the inventory, MAS, SCG, boundary graph, conformance and access summary were checked for the stamp in their source; the deploy's gate over REPORTS checks their output.

🤖 Generated with Claude Code

sam-aydlette and others added 30 commits November 5, 2026 14:32
…ence-only history, period activity

VER-RPT-VDT: carry the historical rating and rating history (providerExtensions;
the schema has no field), take the next reduction from a decision's new
planned_reduction before the remediation deadline, populate supplementary
information from the decision, explain overdue status rather than the
suppression, and date a record from a decision that names it (the manual ZAP
scans behind DEC-039/040).

VER-RPT-AVI: a .checkov.yaml suppression resolves to the resources its decision
names, so its IRV is decided from the inventory; suppressed checks are dated by
their decision.

VER-TFR-MRH: a record that leaves the VDR is Remediated only from evidence (a
closed decision, a false-positive suppression, or its scanner running again);
otherwise it is recorded as no longer reported with the reason and no final
disposition. Closed decisions are backfilled.

VER-RPT-PER: each VDT/AVI carries an activity summary for its period, including
scans with times and the runtime V&V run, linked.

Ratings, deadlines, overdue status and blocking are unchanged on the live inputs.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ed rules

The vendored ksi-catalog.json predated FedRAMP's 2026-10-13 mapping revision:
40 of 46 KSIs mapped fewer (or different) controls than the pinned
Consolidated Rules. scripts/sync-ksi-catalog.py now derives the KSI block
from the pinned rules (Class C statements where a KSI varies by class,
titles from the NIST Rev5 catalog), and a test fails while they differ.

Within the Moderate baseline no implementation status changes; SC-45 and
SC-45.1 become shared. The projection reaches 59 controls beyond Moderate
(was 10). s3.rego drops si-12, which no KSI maps any more. The SSP parity
test runs on the catalog as it stood at the SSP's retirement.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every applicable 20x Class C rule, every KSI and every Rev5 Moderate
control, each no stronger than what backs it; semver versioning; consumers
moved to the schema shape.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…eploy

SDR-CSO-MTD Implemented; SDR-CSO-FRR, SDR-CSX-KSI, FRC-CSX-VVK and
FRC-CSX-VVR gaps restated for the post-deploy state.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…notifications and deadline checks

Every change now carries an explicit significant-change evaluation (SCN-Type,
SCN-Rationale, and SCN-Record for adaptive/transformative/emergency); a
missing tag fails instead of defaulting to routine-recurring, and class-change
blocks. The deploy appends every merged evaluation to a signed, append-only
log and publishes every register row in FedRAMP's SCN schema with a Markdown
view and an Atom feed, listed in the CPO. Business-day deadlines (US federal
holidays) are read from the vendored rules; a scheduled workflow fails on
anything overdue. Notification is publication only; nothing is sent.

SCN-Type: not-significant
SCN-Rationale: change-management process and compliance reporting only; no component, boundary, data flow or access change to the system.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ard)

SCN-Type: not-significant
SCN-Rationale: input validation in the new SCN log builder; no system component, boundary or access change.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…torical notices

The record describes the system after a successful deploy of this commit: the
deploy builds and publishes every notice (any failure fails the deploy; the
round-trip check compares the served index and feed), so the mechanism is in
force. SCN-2026-001..005 were published after their deadlines and are named.

SCN-Type: not-significant
SCN-Rationale: rule-record text and a test only; no component, boundary, data flow or access change.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…/NRI triggers

- data/incidents.json + schema + fail-closed loader (scripts/_incidents.py):
  every incident carries its FedRAMP reportability evaluation and reason;
  default N5 until an IEC-CSO-EFI estimate (checked against the rule's rubric).
- scripts/build-incident-report.py renders Initial/Ongoing/Final reports in
  FedRAMP's incident report schema (validated before write) and the incident
  log; Class C clocks and the rating scale are read from the pinned rules.
- policies/triage.rego incident_triggers: VER-TFR-IRI and VER-TFR-NRI, with
  thresholds parsed from the rule text (fail closed on a changed text).
- Reconciliation invariant (m): an open incident's findings are in the VDR.
- incident-reports.yml publishes unattended (publish-only; nothing is sent);
  bootstrap role scoped to .well-known/incident-*; the deploy also publishes;
  watchdog metric and alarm for overdue reports; freshness entry.
- Dated exercise (2026-11-05, exercise: true): IIR -> 2x OIR -> FIR, plus a
  not-reportable evaluation; run end to end in tests.
- FRR records: EFR, DPR, EFI, VER-TFR-IRI, VER-TFR-NRI Implemented; IIR, OIR,
  FIR, AIR Partially Implemented (publish-only; bootstrap apply pending).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…s state the deployed system

The deploy fails before applying anything unless AWS_INCIDENT_REPORTS_ROLE_ARN
names the bootstrap incident-reports role and the incident-reports environment
deploys only from main (scripts/check-incident-publisher.py; known-bad fixture
with the variable missing). After the deploy job, the deploy calls
incident-reports.yml (now also workflow_call) on its own role, passing only the
CloudFront secret, so a green deploy proves the publisher publishes.

IEC-CSO-IIR/OIR/FIR/AIR gaps now state only what holds after a successful
deploy: published, not delivered to FedRAMP or agencies, by the operator's
decision.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The inbox is simulated for this proof of concept by the operator's decision:
nothing is received or sent. scripts/_fsi.py is the inbox logic
(classification, designators, Class C deadline, register, header-safe
acknowledgement sent once, paging). Every deploy runs it over committed
synthetic messages before the SDR build, fails on any unexpected outcome, and
publishes the signed, dated result as fedramp-inbox-simulation.json, labelled
as a simulation. The published contact says so and points to LinkedIn; the
AFC-CSO-NOC check applies to that contact. AFC records are Implemented with
simulated {counterparty: mail}.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…e SDR

Every deploy and every unattended publish now "delivers" each incident report
the log lists to a simulated FedRAMP and agency recipient
(scripts/deliver-incident-reports.py): read back, checked against the
published sha256 and FedRAMP's schema, and recorded in the signed delivery log
/.well-known/incident-deliveries.json. Nothing is sent. A failed delivery
stops publication.

The deploy job assumes the incident publisher's role before it builds the SDR
(scripts/check-incident-role.py), so a deploy whose publisher role does not
work stops before publishing records that claim it. The role's trust gains the
reviewer-gated deploy environment (bootstrap apply).

IEC-CSO-IIR/OIR/FIR/AIR are Implemented with simulated {incident-recipient}.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…y; SDR carries it; cited tests must pass

Totally Real Assessment Services, LLC (simulated) runs as a reusable workflow
before the deploy job, works from the generated assessor guide, re-runs every
cited check, and signs one set under its own workflow identity, which
infrastructure/keys/assessor-identity.json pins. The SDR build requires and
verifies the set. The unit-test job fails when a test cited as rule evidence
did not pass (FRC-CSX-VVR).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ds in allocated blocks

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ence-only history, period activity

VER-RPT-VDT: carry the historical rating and rating history (providerExtensions;
the schema has no field), take the next reduction from a decision's new
planned_reduction before the remediation deadline, populate supplementary
information from the decision, explain overdue status rather than the
suppression, and date a record from a decision that names it (the manual ZAP
scans behind DEC-039/040).

VER-RPT-AVI: a .checkov.yaml suppression resolves to the resources its decision
names, so its IRV is decided from the inventory; suppressed checks are dated by
their decision.

VER-TFR-MRH: a record that leaves the VDR is Remediated only from evidence (a
closed decision, a false-positive suppression, or its scanner running again);
otherwise it is recorded as no longer reported with the reason and no final
disposition. Closed decisions are backfilled.

VER-RPT-PER: each VDT/AVI carries an activity summary for its period, including
scans with times and the runtime V&V run, linked.

Ratings, deadlines, overdue status and blocking are unchanged on the live inputs.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… dates for DEC-007/009/010 from git history

The CKV_AWS_117 skip is global, so DEC-010 covers every Lambda function; it named
a resource no inventory component carries. The inline skips for CKV_AWS_68,
CKV_AWS_86/310 and CKV_AWS_117 were added on 2025-06-14 (commit f453e9d), so each
was detected by then. AI-drafted for the operator to confirm.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…VDT (VDR-CSO-AKE, VDR-TFR-KEV)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…d disclosure per finding

policies/triage.rego:
- N0 for a fully mitigated finding, read from VER-EVA-EPA's scale (VER-EVA-EPA;
  REQ-PAI-1). N0 has no timeframe.
- Groups (VER-EVA-GRV): a finding's group (CVE, else decision, else check) is
  evaluated at its most exposed open member, on the clock of its earliest
  member; the group's deadline binds every open member (known-bad fixture: a
  new instance of an old group is past the group's deadline and blocks).
- VER-TFR-MAV "will not be": a decision's planned remediation past day 192
  accepts the finding now (accepted_basis).
- VER-TFR-EVU: evaluation_late when first evaluated more than the rules'
  window after detection, or still on a default after it.

scripts/_evaluation.py: the evaluation record per finding (criticality,
reachability, exploitability, detectability, prevalence, privilege, proximate
findings, known threats, automatable, exploitation); the disclosure
determination and the redaction it drives (VER-RPT-NID/RPD). A decision record
may carry automatable {value, evidence}; 'no' without evidence is refused.

The golden triage report is regenerated: the only changed values are the three
fully mitigated findings (N1 -> N0); every other change is an added field.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…oy (MAS-CSO-IIR/FLO/MDI/TPR, FRC-CSX-MAS)

- Inventory: the app's browser origins (jsDelivr, Google Cloud Storage) and the
  VDR's remaining feeds (FIRST EPSS, Grype database) become external services;
  the CloudTrail trail joins the logging KSIs' evidence domain.
- scripts/build-mas.py publishes /.well-known/minimum-assessment-scope.json
  (+ .md, signed, freshness-governed): information resources, every CSP
  off-origin host placed or out of scope with a reason, flows with a
  high-water-mark security category, a metadata inventory, the third-party
  records with FedRAMP's four fields, and the KSI-to-scope application. It
  fails the deploy, before the SDR, on any gap.
- Classification: 17 new flows covering every external service and data store;
  an app-delivery group (and its service in the service model).
- Third-party records: the four MAS-CSO-TPR fields on all 11; decision fields
  AI-drafted for the operator to confirm; a decided record without them fails.
- scripts/_tpr.py: the CPO's thirdPartyInformationResources, for the CPO hook.
- Five rule records flip to Implemented with resolving evidence.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
providerExtensions.fedRampId, read from data/cds/offering.json fedramp_id (the
field _offering.fedramp_id() reads on feat/sdr-schema). A missing ID stops the
build. Adding these files to check-fedramp-id.py's list is left for integration.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ng standard; DEC-165..167

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…sclosure (DEC-182) policies

The drafted per-check CVSS vectors are adopted by a decision record, and each
names it: a decided vector whose record is missing or not in force stops the
build (known-bad fixture). Configuration findings now get PAIN from the
CVSS-Environmental derivation and LEV from Figure 7 instead of the defaults
(VER-EVA-ELX, VER-EVA-EPA). DEC-180..182 are AI-drafted for Sam's confirmation.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…, CORS on /.well-known/

- availability-probe.yml + scripts/availability-probe.py: a monitor on GitHub
  Actions (outside AWS) probes the site, trust center, /.well-known/, the Silk
  Reeling app and its sign-in every 15 minutes, keeps 35 days, derives
  incidents, and publishes JSON + Markdown to the availability-status branch;
  an ongoing incident fails the run. The deploy fails when that feed is missing
  or stale, and availability.json is now built from it, never from deploys.
- CloudFront standard logging v2 (bootstrap/trust-access-logs.tf) with a field
  allowlist that drops IP, cookies, query string, user agent and referrer;
  raw records expire after 90 days (logging.tf). The deploy fails unless the
  delivery exists within the allowlist, and publishes monthly aggregates for
  /.well-known/ and /trust/ only (access-summary.json, 13 months).
- /.well-known/* gets its own response-headers policy with read-only CORS;
  the deploy checks CORS, JSON Content-Type and nosniff live.
- DEC-160 (AI-drafted for the operator): what is logged, and a party's own
  records on request.
- CDS-CSO-AVR, CDS-TRC-USH, CDS-TRC-ACL flip to Implemented.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…er-service crypto modules

The Secure Configuration Guide is generated on every deploy from one source
(data/secure-configuration.json) and published signed as
/.well-known/secure-configuration-guide.json and .md: each top-level and
privileged account with how it is named, accessed, configured, operated and
decommissioned; each setting with who can operate it, the recommended value,
its implication and the API that reads and changes it; a version and release
history whose defaults digest the build enforces. It states the system as it
is: virtual TOTP MFA on root and the operator user, an accepted finding
(DEC-150, extending DEC-025).

scripts/check-secure-config.py reads the live settings read-only before the
SDR is built, compares them with the guide and publishes the export as
/.well-known/security-settings.json; an uncovered deviation or an unreadable
setting fails the deploy. The deploy role's reads are a new bootstrap policy
(infrastructure/bootstrap/deploy-config-reads.tf), applied by the operator.

The guide also carries the cryptographic module per service, joined with the
inventory (the build fails on an unmapped component type), citing the KMS
HSM's CMVP certificate (#4884) and stating where AWS publishes none (DEC-152).

Flips SCG-CSO-AUP/PUB/RSC/SDF, SCG-ENH-API/CMP/EXP/MRG/VRH and
CMU-CSO-CMD/UVM/CAT to Implemented. Decision records DEC-150..152 and the
guide's text are AI-drafted for the operator to confirm.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ished documents; lossless human views with round-trip check

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
# Conflicts:
#	data/decisions.json
#	infrastructure/schemas/freshness-config.json
# Conflicts:
#	.github/workflows/deploy-with-opa.yml
#	data/decisions.json
#	tests/test_decisions.py
…site root resolves to its page

VDR-CSO-FAV: a detect-and-recover failure recorded in the run, or a finding
still on a default after the VER-TFR-EVU window, becomes a finding with source
'process' (tracking ID, clock, rating, deadline), remediated when the check
passes. Runs a block-mode check stopped are listed in the next report's
process_failures (workflow and times; no run IDs or URLs), from a new
'Collect failed pipeline runs' step in the deploy and the nightly.

VER-EVA-EPA: impact_rationale states the likely agency effect and what it rests
on (the decided vector's rationale, the advisory's vector, or the severity
default), and the VDT's potentialAgencyImpact carries it. The VDT's
providerExtensions carry the evaluation record, group, disclosure and
acceptance basis.

VER-EVA-EIR: a DAST finding on a site root (https://host) resolves to the root
page instead of the fail-safe.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
sam-aydlette and others added 27 commits November 5, 2026 17:11
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…trust-scope-assessor

Conflicts: deploy-with-opa.yml (edge steps kept before group 2's renamed CDS
step; the folded-in COP dropped from the trust-scope lists; group 2's report
signing and redaction entries kept), offering.json (the SCG fork's module
validations in group 2's layout), decisions.json (union by ID),
test_frr_records.py (theirs). DEC-150 now names the operators group, not an
IAM user, so the new no-IAM-user-name test holds; IRP evidence follows the
renamed CDS step.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…es for every machine component type

One Rego source now gates the plan and the live account for IAM roles,
inline and group policies, the GitHub OIDC provider and trust, the Cognito
user pool, KMS keys, the ACM certificate, DNSSEC, log groups, CloudTrail,
alarms, schedules, the API stage and routes, Lambda functions and their
deployed code package, SNS, SQS and DynamoDB. Rules read Terraform
attribute names under values, which the runtime emitter writes from the
live APIs; an attribute computed at apply time is deferred to the runtime
check, never assumed good. New parameters (OIDC audience and subject
prefix, password floor, log retention floor, protected API prefixes,
certificate horizon) are governed data in config/data.json and required by
the configuration guard. Known-bad plan and runtime fixtures cover each
rule family.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…e defers to unread actions

The Wasm host has no time builtins, so the certificate rule reads a
days_to_expiry the reader computes instead of parsing NotAfter. An alarm whose
actions could not be read is reported once, as the read error.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Plan results now become validations of IAM roles and policies, log groups,
keys, the user pool, topics, queues, schedules, alarms and the trail, joined
by exact plan address first (resource_reports carries it), then by a unique
Terraform name; ambiguous names credit nobody. API stages and routes fold into
their API. Function components carry the deployed code hash and npm/PyPI
components name the function they ship in, for the runtime integrity check.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
history.js summarizes each run (signal_id, emitted_at, validations, failed,
unassessed, divergence) and carries previous summaries forward only from a
signal that verifies against the emitter's own key; otherwise the history
restarts and says why. 400 entries, schema-bounded.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…licy

describers.js reads IAM roles, policies, groups and the OIDC provider,
functions (configuration and deployment-package integrity), log groups, the
trail, alarms, schedules, topics, queues, the lock table, the user pool, KMS
keys, the HTTP API, the DNS zone and the certificate, and shapes each as the
Terraform resource the deploy gate's Rego evaluates. A denied call becomes
read_errors (unassessed), never a pass. index.js wires the readers in and
publishes run_history inside the signed signal.

The lambda_opa role gains read-only Get/Describe/List actions, each scoped
by ARN to this system's resources; tag:GetResources is the only action on
"*" (no resource-level support). tests/test_runtime_grants.py ties every
SDK call to a grant and every grant to a call. Timeout 60s -> 180s.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every machine component of the inventory needs an assessed validation in a
verified runtime signal no older than the governed window; every component
type must be classified in runtime-vv-config.json, which also records the
likely-to-drift set (VDR-TFR-PDD). The nightly reconciliation runs it against
the published runtime signal; a missing or unverifiable signal fails it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The VDR build carries the emitter's signed run history (from a verified
signal only) in methodology.runtime_verification.run_history; the VDT/AVI
activity summary and the monthly report count every run in the period from it,
not only the latest.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…orting and detection Implemented; VVK at 43 of 46 KSIs

VDR-TFR-MVX, -PDD, -PSD; VER-RPT-PER; VDR-CSO-DET Implemented with their
evidence. FRC-CSX-VVK stays Partially Implemented: SCR-MIT and SCR-MON have one
automated method and CED-RAT none; closing it needs an SDR counting change or
the operator's decision.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…he assessor's summary; FedRAMP ID on the trust-center files

Wire group 5's four CPO fields into build-cpo.py, one small function each:
serviceProperties.trustCenter, serviceProperties.secureConfigurationGuidance,
thirdPartyInformationResources (scripts/_tpr.py) and the simulated
assessor's overall summary, quoted verbatim from its verified signed set
under providerExtensions.independentAssessment. The CPO refuses to build
without the set; the deploy and the daily package maintainer pass it (the
maintainer also passes it to the SDR rebuild, which previously dropped the
assessor results). The package submission now refuses a CPO missing any of
the four (known-bad fixture).

Stamp the offering's FedRAMP ID at top level on trust-center.json,
trust-manifest.json, availability.json and the independent availability
feed, policy-index.json, assessor-guide.json, simulated-assessment.json,
agency-access-exercise.json and frr-evidence.json, add them to the gate's
REPORTS, make a named site file without one fail (known-bad fixture), and
run the gate once the last listed file is built, before publication.

Flip CPO-CSO-OSA (simulated assessor), CPO-CSO-OVR, CDS-CSO-PUB (UEI
simulated sam-gov), FRC-CSO-JSN and FRC-CSO-PKG (simulated fedramp) to
Implemented. CDS-CSO-FID stays Partially Implemented for integration.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflicts resolved keeping both sides:
- offering.json: the schema-form CPO values, plus the simulated FedRAMP
  Security Inbox contact (no personal email address is restored).
- build-cpo.py: the CPO stays in FedRAMP's schema; the inbox contact and the
  SCN notification statement move to providerExtensions, and the SCN, incident
  and inbox artifacts join the package's additional repositories.
- build-ocr.py: the schema-form OCR keeps its incident list and gains the
  validated incident register and the attestation of none.
- freshness-config.json: all entries from both sides; the retired CDS
  overview package stays retired.
- boundary-classification.json: the incident publisher flow is renumbered Z.
- deploy-with-opa.yml: both preflights, both post-deploy proof jobs, every
  artifact signed, gated and published; SCN and FRC steps run before the SDR.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflicts resolved keeping both sides:
- frr-records.json and decisions.json merged record by record: each record
  from the branch that changed it; FRC-CSX-VVK from the detection work.
- triage.rego: the grouped decide (base_decisions plus grouping) with the
  incident triggers and errors appended after decisions.
- reconcile.py: both invariants kept. Incident findings stay (m), as the
  gate-mode register and the published report name it; runtime V&V coverage
  becomes (n) and still always blocks.
- deploy-with-opa.yml and evidence-nightly.yml: incident reports, the
  non-machine V&V check and the monthly report all built, signed, gated and
  published; the nightly rebuilds human-readable views for the two new
  VDR artifacts.
- non-machine-resources.json: the FedRAMP Security Inbox runbook joins the
  quarterly verification register.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every builder that writes a freshness-governed JSON artifact stamps the
offering's FedRAMP ID through _offering.stamp_fedramp_id: in
providerExtensions for FedRAMP-schema reports, at the top level for site
files. check-fedramp-id.py covers the VDR family, the KSI signal, SCN and
incident files, the inbox simulation, MAS, SCG, settings, the access
summary, boundary map, reconcile report and conformance, and a test fails
when a governed artifact is neither checked nor exempt with a reason. The
nightly and the incident publisher run the gate on what they republish.
SCN notifications and incident reports are written through
write_validated (FRC-CSO-JSN).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… (FRC-CSX-VVK)

build-sdr.py counts a third automated method for a KSI with npm or PyPI
components: the bound VDR's signature verifies against the deploy or
nightly identity, it is within its window, and every Grype feed passed the
detection gate. KSI-CED-RAT counts the operator's training attestation
(DEC-192) while it is in force and under a year old. A KSI short of two
methods without an in-force accepted decision is a warning in the deploy
(which sees the pre-deploy runtime signal) and fails the package
maintainer, which runs after the deploy and daily. DEC-191 accepts
KSI-CED-RAT's single method. Both decisions are AI-drafted for the
operator to confirm.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
aws_iam_group_policy, aws_iam_role_policy_attachment and
aws_iam_group_policy_attachment become iam_policy components, one per
attachment block however many instances its for_each makes, with attached
policies listed by name only. Scanner findings that name these blocks now
resolve to a component instead of vdr.unresolved-component. Address
resolution prefers an exact resource type before the prefix match, so an
attachment named like its group resolves to the attachment.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…SCN and incident JSON

CDS-CSO-FID cites the coverage test, its known-bad fixtures and the
nightly and incident-publisher gate steps. FRC-CSX-VVK describes the four
methods, states that KSI-CED-RAT has one with the accepted finding
DEC-191, and cites the SDR tests, the maintainer step and both decisions.
FRC-CSO-JSN now cites the SCN and incident schema tests.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…uard)

Keeps the expression out of the shell script body.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…its parser

Both merged branches added the option, and argparse refuses a second
declaration, so the OCR step would have stopped the deploy. A test runs
--help on every argparse script a workflow calls, with a known-bad fixture.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…on gate checks for them

The vulnerability history backfills closed decisions with the resources
they name, and the trust center and VDR publish decision scopes; the
register names IAM users as aws-iam-user::<name>. Each published copy now
carries aws-iam-user::NAME-WITHHELD, and check-published-redaction.py fails
on any remaining IAM user name, with a known-bad fixture.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ed before publishing

After apply, the deploy stages its signed inventory under
deploy-staging/<sha>/ and invokes the runtime emitter synchronously with
{stage: <sha>}. The emitter accepts only a 40-hex SHA, re-validates the
deployed system against that inventory, and writes its signed signal
beside it (a new PutObject grant on that key only). The deploy reads it
back, deletes the staging objects, and checks it re-validated this
commit. The VDR and SDR use that signal; build-sdr.py --runtime-commit
counts the runtime method only for it, and --require-two-methods fails
the deploy before anything is published. The FRC-CSX-VVK record now
describes the deploy holding the requirement.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… SDR requires two methods

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… one

The nightly beacon names its run by workflow and commit instead of a run
URL; the KSI signal's provenance drops builder.run_id (source.commit and
builder.id identify the build); the SLSA provenance omits
runDetails.metadata.invocationId, which SLSA v1.0 does not require
(BuildMetadata has no required field). check-published-redaction.py fails
on a run URL or a numeric run_id, with known-bad fixtures.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
security.txt and the KSI signal's operator_contact named a personal mailbox.
Both now point at https://www.linkedin.com/in/sa2/, already linked from every
page. A test keeps a personal mailbox out of both.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The operator classed the combined change as adaptive: two unattended IAM roles,
new runtime-emitter grants, CloudFront access logging and a DAST scanner user.
Adds the register row and the impact analysis (AI-drafted for review). The
deadline tests used SCN-2026-007 as an unregistered ID; they now use
SCN-2026-098.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@sam-aydlette
sam-aydlette merged commit 59eb6d8 into feat/frr-records Nov 6, 2026
2 checks passed
@sam-aydlette
sam-aydlette deleted the feat/integration branch November 6, 2026 12:37
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant