Skip to content

v0.10.0 — critical interaction and actionability protocol upgrade

Choose a tag to compare

View all tags
@openleek openleek released this 11 Jun 15:20
· 17 commits to main since this release
v0.10.0
3e7557e
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Highlights

  • Added a schema-backed critical interaction routing gate for blocking clarification, progressive follow-up, workflow handoff, follow-up continuation, and explicit waiver.
  • Strengthened Mira's human / model / protocol boundary: Mira remains a research protocol, model output is not evidence, and action prompts route through research posture rather than trade execution.
  • Added the marginal buyer payoff bridge and tightened actionability discipline for buy/add/sell/trim style prompts.
  • Made progressive follow-up a workflow control point across routing, monitoring, event delta, thesis update, SEC/earnings work, position review, portfolio review, decision-quality review, view continuity, and long-term thesis workflows.
  • Kept user-facing interaction natural-language while preserving machine-checkable routing state internally.
  • Expanded behavior eval coverage for critical interaction, workflow handoff, blocking clarification, model/persona boundaries, quick-map follow-up, and SEC supplement routing.
  • Improved repo validation by excluding local tool state directories such as .claude/, private/, and local/ from full-repo product checks.

Why This Release Matters

This release moves Mira closer to the intended collaboration model: not pure Q&A, not autonomous execution, but a research workflow where the model answers, exposes uncertainty, and uses one high-value interaction to move the user into a better next step.

Validation

  • python3 scripts/run_quality_gate.py passed on 3e7557e7c8c1fc2b86286efb4d9db20e3a92d1d3.
  • Required checks passed: py_compile, repo_validation_contract_tests, repo_validation, and release_gate.
  • Additional PR validation for #87 included python3 scripts/test_routing_schema.py and behavior eval report-only with 28 passed, 0 failed, 8 missing transcripts.

Known Follow-Up

  • Record transcripts for the 8 new behavior eval cases so just eval-strict can become the release-grade behavior gate for this interaction layer.
Assets 2
Loading