Skip to content

feat(cli): expose Archera comparison filters already available in MCP #2264

Closed
#2265
Closed
feat(cli): expose Archera comparison filters already available in MCP#2264
#2265
Assignees
cristim
Labels
effort/sHoursimpact/fewLimited audiencepriority/p2Backlog-worthyseverity/lowMinor harmtriagedItem has been triagedtype/featNew capabilityurgency/this-quarterWithin the quarter

Description

@cristim
cristim
opened on Oct 18, 2026

Summary

The explicit archera-comparison CLI command currently sends only ComparisonRequest.PlanID. The MCP comparison tool already exposes line-item, contract-term and payment filters supported by the same shared insurance client.

Expected behavior

Add --line-item-ids, --contract-terms and --payment-options to the existing read-only command, with the MCP bounds (200 UUIDs, supported term count, three payment options) and exact shared enums. Accept repeated flags and comma-separated entries. Omission preserves existing behavior; explicit empty/invalid or over-limit input fails before any vendor request. Error output must not echo user values or secret material.

Scope and existing caller

This is an explicit operator command narrowing a vendor comparison. Use the already pinned pkg/insurance ComparisonRequest; no dependency bump, matcher or new abstraction is required. IDs belong to Archera, not to platform recommendation rows. This does not claim per-recommendation association.

Verification

Drive the real root command and real insurance.Client with intercepted synthetic transport. Assert GET path/auth and repeated query values, omission controls, malformed/empty/repeated-empty and exact bound cases making zero HTTP requests, output redaction, help without calls, and flag state reset across repeated executions. Preserve existing rendering and disclosures. Run pinned build, unit/race, vet/lint, normal hooks, two independent Sol High reviewers each two passes on current head, and CI.

References and triage

Companion to LeanerCloud/cloud-commitments-platform#799 and #786. Existing CLI #2156/PR #2166 delivered plan-wide comparison; #2171 and #2259 track different issues. No live vendor calls, purchase, scheduling, deployment or credential changes.

Activity

  1. cristim
    self-assigned this
    on Oct 18, 2026
  2. cristim
    added
    triagedItem has been triaged
    priority/p2Backlog-worthy
    severity/lowMinor harm
    urgency/this-quarterWithin the quarter
    impact/fewLimited audience
    effort/sHours
    type/featNew capability
    on Oct 18, 2026

  3. cristim commented on Oct 18, 2026

    @cristim
    cristim
    on Oct 18, 2026
    MemberAuthor

    Claiming: resume-cc-audit, scoped CLI/MCP Archera comparison-filter parity under the owner-authorized completion goal. Plan review first, then an isolated worktree; existing checkpoints and unrelated CLI issues remain untouched. [resume-cc-audit]

  4. cristim
    closed this as completedin #2265on Oct 18, 2026

  5. cristim commented on Oct 18, 2026

    @cristim
    cristim
    on Oct 18, 2026
    MemberAuthor

    Session: resume-cc-audit

    Merged #2265 normally as 021f16279c119effc261bbc08e560c27f67594a5. Its full tree 9026b1296c5c46cda90d336ae73abf077aeb9c2e exactly matches the reviewed and tested head 28574f7d54be0e43b38a374b88bc31a5b8940a06.

    Verified:

    • Two independent Sol High reviewers each completed two adversarial passes, closed the findings and approved the exact head. Reviewer A independently reproduced the real root-command/shared-client focused race run: 31 top-level tests, 206 including subtests.
    • Seven positive filter scenarios failed on the parent with unknown flags; the omission control passed. Final full race suite passed 471 top-level tests. Build, vet, pinned lint and normal hooks passed.
    • Completed CodeRabbit review covered the exact head with no actionable comments. Required PR CI passed.
    • Post-merge CLI help shows all three filters and the actual 74-term limit. Post-merge main runs CI and pre-commit both passed.
    • Report-first, exact-branch gated cleanup reclaimed the clean worktree and local/remote feature refs under reclaim and push locks. The session-built binary was preserved separately; detached and ambiguous recovery work remained untouched. Final report: zero eligible items.

    This completes the filter-parity issue. Tests use intercepted synthetic responses; no live vendor call, purchase, installation or deployment was performed. Real recorded-response acceptance remains tracked by platform #798; recommendation integration remains platform #799.

    Follow-up issues filed: none for this filter change, confirmed against the follow-up checklist.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

effort/sHoursimpact/fewLimited audiencepriority/p2Backlog-worthyseverity/lowMinor harmtriagedItem has been triagedtype/featNew capabilityurgency/this-quarterWithin the quarter

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions