Issues

• ci: the code-lane gate takes ~8.5 min, almost all pytest; split serial tests, cut CLI cold start, consider sharding

  ModelMirrorAI/fedcourtsai#2528 · Filter by author modelmirror-agent opened on Oct 18, 2026

• interim: the salience reserve and moment minting don't read the one-row-per-number skip set

  data-quality
  ModelMirrorAI/fedcourtsai#2521 · Filter by author modelmirror-agent opened on Oct 17, 2026 · Provisioning & document coverage

• backtest: read replay cells' flags.json through the normaliser, so a mis-shaped file isn't recorded as unreadable

  ModelMirrorAI/fedcourtsai#2515 · Filter by author modelmirror-agent opened on Oct 17, 2026 · Back-test program

• outcome: a re-firing mint rewrites event.yaml's caption, so the re-docket check loses its committed witness

  ModelMirrorAI/fedcourtsai#2514 · Filter by author modelmirror-agent opened on Oct 17, 2026 · Provisioning & document coverage

• documents: the contact scrub's email pattern is quadratic on long runs, and the street pattern withholds an Ohio State reporter citation

  ModelMirrorAI/fedcourtsai#2513 · Filter by author modelmirror-agent opened on Oct 17, 2026 · Provisioning & document coverage

• summaries: keep documents the contact scrub flags with residuals out of the public summary-stage artifact

  data-quality
  ModelMirrorAI/fedcourtsai#2512 · Filter by author modelmirror-agent opened on Oct 17, 2026 · Provisioning & document coverage

• security: an injected cell in a prod-bound job can mint an OIDC token the production read-write role and the Claude GitHub App accept

  ModelMirrorAI/fedcourtsai#2511 · Filter by author modelmirror-agent opened on Oct 17, 2026 · Engine auth: OIDC / workload identity
  1

• Automate the promotion's sync → test → prepare sequence; make the contexts check on-demand

  ModelMirrorAI/fedcourtsai#2502 · Filter by author modelmirror-agent opened on Oct 17, 2026

• Process registry: stamped rows can change after the promotion, and the remaining golden pins break main at the first label stamp

  ModelMirrorAI/fedcourtsai#2500 · Filter by author modelmirror-agent opened on Oct 17, 2026

• metrics: pre-register a self-interest comparison — predictors forecasting litigation against their own developer

  ModelMirrorAI/fedcourtsai#2458 · Filter by author modelmirror-agent opened on Oct 15, 2026 · Lower-court slices: AI cases and the D.C. Circuit

• resolution: deterministic district ruling resolver for the watchlist's trial moments

  ModelMirrorAI/fedcourtsai#2457 · Filter by author modelmirror-agent opened on Oct 15, 2026 · Lower-court slices: AI cases and the D.C. Circuit

• district: pre-register a court-agnostic trial stage — moments and outcomes for motions, judgment and settlement

  ModelMirrorAI/fedcourtsai#2456 · Filter by author modelmirror-agent opened on Oct 15, 2026 · Lower-court slices: AI cases and the D.C. Circuit

• corpus: ingest the AI watchlist's dockets, district courts included, without admitting their courts

  ModelMirrorAI/fedcourtsai#2455 · Filter by author modelmirror-agent opened on Oct 15, 2026 · Lower-court slices: AI cases and the D.C. Circuit

• ai-cases: commit the AI-case watchlist and its rule, and tag its cases across courts

  ModelMirrorAI/fedcourtsai#2454 · Filter by author modelmirror-agent opened on Oct 15, 2026 · Lower-court slices: AI cases and the D.C. Circuit

• ai-cases: AI-case watchlist slice — tracking (milestone 2 in docs/milestones.md)

  ModelMirrorAI/fedcourtsai#2453 · Filter by author modelmirror-agent opened on Oct 15, 2026 · Lower-court slices: AI cases and the D.C. Circuit

• feeds: an AI-cases feed — just the facts on watchlist dockets, predictions as opt-in enrichment

  ModelMirrorAI/fedcourtsai#2452 · Filter by author modelmirror-agent opened on Oct 15, 2026 · Public update feeds

• feeds: optional public email updates — weekly SCOTUS update and AI-cases feed — tracking

  ModelMirrorAI/fedcourtsai#2449 · Filter by author modelmirror-agent opened on Oct 15, 2026 · Public update feeds

• feedback: make agent flags triageable — never fail a cell, collapse re-reports, mark tracked themes (harness now; prompt in proc-v10)

  ModelMirrorAI/fedcourtsai#2442 · Filter by author modelmirror opened on Oct 15, 2026 · Agent observability

• predict: gemini-baseline's retrieval.md says no retrieval while its harness log records corpus queries

  ModelMirrorAI/fedcourtsai#2425 · Filter by author modelmirror opened on Oct 14, 2026 · Agent observability
  1

• predict prompt: a cert "distribution" cell reads the latest snapshot uncut, but predict.md says it is forecast at first distribution

  ModelMirrorAI/fedcourtsai#2423 · Filter by author modelmirror opened on Oct 14, 2026 · Next re-bless: proc-v10 / sal-v5 candidates

• corpus/outcomes: a reassigned SCOTUS docket number scores the arrival cell against another petition (scotus/9026000221)

  data-quality
  ModelMirrorAI/fedcourtsai#2422 · Filter by author modelmirror opened on Oct 14, 2026 · Provisioning & document coverage
  3

• base rates: decide the interim pool's lookback window for OT2027 (shared 10-Term window vs a shorter interim window)

  ModelMirrorAI/fedcourtsai#2396 · Filter by author modelmirror opened on Oct 12, 2026 · Waiting on data or a decision
  1

• dev: make the Claude app the primary development surface — readiness tracker

  ModelMirrorAI/fedcourtsai#2391 · Filter by author modelmirror opened on Oct 12, 2026 · CI & engine hygiene
  1

• docs: re-anchor budget.md and milestones.md after the moves to GPT-6.1 Sol and Gemini 4 Argon

  ModelMirrorAI/fedcourtsai#2390 · Filter by author modelmirror opened on Oct 12, 2026 · Waiting on data or a decision

• promotion: automate the integration-gate dispatch at the ready promotion head

  ModelMirrorAI/fedcourtsai#2389 · Filter by author modelmirror opened on Oct 12, 2026 · CI & engine hygiene
  1