Issues

• Add portable evals for the six agents exempted from eval pairing

  evals & fixtures
  Bug

  cosai-oasis/secure-ai-tooling#567 · Filter by author davidlabianca opened on Oct 2, 2026

• ADR branch-placement and landing-granularity reconciliation for an architect pass (ADR-002 / ADR-027 / ADR-034)

  adr
  Task

  cosai-oasis/secure-ai-tooling#563 · Filter by author davidlabianca opened on Oct 1, 2026
  

• Advance MITRE ATLAS from 5.0.1 to 2026.09 (ADR-027 D10)

  frameworks
  Task

  cosai-oasis/secure-ai-tooling#562 · Filter by author davidlabianca opened on Oct 1, 2026
  

• Controls guidance (P4): port three existing controls

  control
  Task

  cosai-oasis/secure-ai-tooling#560 · Filter by author davidlabianca opened on Sep 30, 2026
  

• Controls guidance (P3): contributor teaching text

  documentation
  Task

  cosai-oasis/secure-ai-tooling#559 · Filter by author davidlabianca opened on Sep 30, 2026
  

• Controls: separate the control statement from implementation guidance

  adrdocumentationtooling
  Feature

  cosai-oasis/secure-ai-tooling#556 · Filter by author davidlabianca opened on Sep 30, 2026
  

• [tests]: regenerate-graphs tests don't pin the exact validator invocation; stale test-file metadata

  testing
  Bug

  cosai-oasis/secure-ai-tooling#554 · Filter by author davidlabianca opened on Sep 30, 2026
  

• [tests]: test_styling_docs_schema_examples.py deep-merges doc examples into the live config, masking examples that are invalid on their own

  testing
  Task

  cosai-oasis/secure-ai-tooling#553 · Filter by author davidlabianca opened on Sep 30, 2026

• [schema]: a whitespace-only decoupled port style passes the schema and emits a classDef with no body

  pre-committooling
  Bug

  cosai-oasis/secure-ai-tooling#552 · Filter by author davidlabianca opened on Sep 30, 2026
  

• ADR-033 D7b: eval fixtures must not carry grading notes, and the shipped fixtures must be retrofitted

  adrevals & fixtures
  Task

  cosai-oasis/secure-ai-tooling#540 · Filter by author davidlabianca opened on Sep 24, 2026
  

• ADR-033 D7a: ground every corpus-dependent eval expectation, and sweep the shipped suites

  adrevals & fixtures
  Task

  cosai-oasis/secure-ai-tooling#539 · Filter by author davidlabianca opened on Sep 24, 2026
  

• [infra]: Semantic Versioning standard for RM et al.

  enhancementinfrastructure
  cosai-oasis/secure-ai-tooling#534 · Filter by author jo5iah opened on Sep 16, 2026

• [Control]: controlAgentMemoryIntegrity

  controlnew-control
  cosai-oasis/secure-ai-tooling#527 · Filter by author jo5iah opened on Sep 6, 2026
  3

  

• [Risk]: riskUnsafeInterAgentPropagation

  new-riskrisk
  cosai-oasis/secure-ai-tooling#526 · Filter by author jo5iah opened on Sep 6, 2026
  2

  

• [Risk]: riskDeceptiveAgentReporting

  new-riskrisk
  cosai-oasis/secure-ai-tooling#525 · Filter by author jo5iah opened on Sep 6, 2026
  10

  

• [Risk]: riskAgentMemoryPoisoning

  new-riskrisk
  cosai-oasis/secure-ai-tooling#524 · Filter by author jo5iah opened on Sep 6, 2026
  3

  

• Reconcile draft-stage hand-composed mapping values with ADR-027 D4

  adragentsskillstooling
  Feature

  cosai-oasis/secure-ai-tooling#522 · Filter by author davidlabianca opened on Sep 1, 2026

• Content review: cross-cutting entries — 4 entries in PR #507 that belong to no single subject area

  controlnew-controlnew-riskrisk
  Task

  cosai-oasis/secure-ai-tooling#521 · Filter by author davidlabianca opened on Aug 31, 2026

• Content review: the classical AI/ML estate — 13 entries in PR #507

  controlnew-controlnew-riskrisk
  Task

  cosai-oasis/secure-ai-tooling#520 · Filter by author davidlabianca opened on Aug 31, 2026

• Content review: prompt injection and context integrity — 4 entries in PR #507

  controlnew-controlnew-riskrisk
  Task

  cosai-oasis/secure-ai-tooling#519 · Filter by author davidlabianca opened on Aug 31, 2026
  1

• Content review: observability, audit and accountability — 7 entries in PR #507

  controlnew-controlnew-riskrisk
  Task

  cosai-oasis/secure-ai-tooling#518 · Filter by author davidlabianca opened on Aug 31, 2026
  6

• Content review: network enforcement — 5 entries in PR #507

  controlnew-controlnew-riskrisk
  Task

  cosai-oasis/secure-ai-tooling#517 · Filter by author davidlabianca opened on Aug 31, 2026

• Content review: isolation and containment — 5 entries in PR #507

  controlnew-controlnew-riskrisk
  Task

  cosai-oasis/secure-ai-tooling#516 · Filter by author davidlabianca opened on Aug 31, 2026
  1

• Content review: tool supply chain — 8 entries in PR #507

  controlnew-controlnew-riskrisk
  Task

  cosai-oasis/secure-ai-tooling#515 · Filter by author davidlabianca opened on Aug 31, 2026
  1

• Content review: authorization and delegation-chain integrity — 10 entries in PR #507

  controlnew-controlnew-riskrisk
  Task

  cosai-oasis/secure-ai-tooling#514 · Filter by author davidlabianca opened on Aug 31, 2026
  2