Issues

• Generated reviewer subagents have no per-persona model/effort: expose the documented frontmatter fields through members.yml

  loop:researchstatus:analyzed
  codenamev/ai-software-architect#67 · Filter by author minerva-sky opened on Sep 19, 2026
  1

• Review findings carry severity but no confidence: add per-finding confidence at emission and a dedup pass at aggregation

  loop:researchstatus:analyzed
  codenamev/ai-software-architect#66 · Filter by author minerva-sky opened on Sep 19, 2026

• CI validates the plugin with hand-rolled checks; claude plugin validate / claude plugin details run offline and know the current spec

  loop:selfstatus:analyzed
  codenamev/ai-software-architect#61 · Filter by author minerva-sky opened on Sep 17, 2026

• The plugin path has no behavioral test: replace the one-prompt claude-smoke job with a claude plugin eval suite (with no-plugin baseline)

  loop:selfstatus:analyzed
  codenamev/ai-software-architect#60 · Filter by author minerva-sky opened on Sep 17, 2026
  1

• Installed .architecture/ silently drifts from the auto-updating plugin: nothing reads framework_version, and the template stamps 1.2.0

  loop:selfstatus:analyzed
  codenamev/ai-software-architect#59 · Filter by author minerva-sky opened on Sep 17, 2026

• dependabot.yml: security updates bypass the groups — add applies-to: security-updates so advisory fixes arrive as one PR, not three

  loop:depsstatus:analyzed
  codenamev/ai-software-architect#58 · Filter by author minerva-sky opened on Sep 15, 2026

• mcp: drop the @types/node devDependency — nothing in the package consumes it, and it generates a Dependabot PR every week

  loop:depsstatus:analyzed
  codenamev/ai-software-architect#57 · Filter by author minerva-sky opened on Sep 15, 2026

• Research digest: spec-kit, BMAD, OpenSpec, MADR, log4brains — 4 integration candidates, 3 settled non-candidates

  loop:researchstatus:analyzed
  codenamev/ai-software-architect#38 · Filter by author minerva-sky opened on Aug 22, 2026
  1

• Reviews emit findings but never learn: record per-finding dispositions to make persona quality measurable

  loop:selfstatus:analyzed
  codenamev/ai-software-architect#37 · Filter by author minerva-sky opened on Aug 20, 2026
  1

• members.yml promises a collaborative phase the orchestrator never runs — implement it as a structured challenge round

  loop:selfstatus:analyzed
  codenamev/ai-software-architect#36 · Filter by author minerva-sky opened on Aug 20, 2026
  1

• Personas are noun taxonomies, not reviewers — add behavioral fields (silent checklist, blocking criteria) to members.yml

  loop:selfstatus:analyzed
  codenamev/ai-software-architect#35 · Filter by author minerva-sky opened on Aug 20, 2026
  1

• After #20: adopt the SDK 1.x API so advertised inputSchema contracts are actually enforced

  loop:depsstatus:analyzed
  codenamev/ai-software-architect#30 · Filter by author minerva-sky opened on Aug 18, 2026

• All CI runtime pins are past end-of-life (Node 20, Node 18, Python 3.9) and engines publishes an EOL floor

  loop:depsstatus:analyzed
  codenamev/ai-software-architect#29 · Filter by author minerva-sky opened on Aug 18, 2026

• configure_pragmatic_mode destroys all comments in config.yml (yaml parse/stringify round-trip)

  loop:qualitystatus:analyzed
  codenamev/ai-software-architect#26 · Filter by author minerva-sky opened on Aug 14, 2026

• Setup fidelity leftovers (ADR-016): hardcoded templates, template-blind create_adr, discarded .coding-assistants content

  loop:qualitystatus:analyzed
  codenamev/ai-software-architect#25 · Filter by author minerva-sky opened on Aug 14, 2026

• Skills-based setup copies framework's own files into target projects

  codenamev/ai-software-architect#9 · Filter by author sgbett opened on Feb 14, 2026
  1