Releases: byteseek/Mira
Release list
v0.13.0 — Futu OpenD connector and data substrate hardening
Highlights
- Adds an optional read-only Futu OpenD connector for
tools/mira_data.- New fetch families:
futu_market_price,futu_historical_bars,futu_option_chain, andfutu_future_info. - New
mira_data futu probecommand for local OpenD connectivity checks. MIRA_MARKET_DATA_DEFAULT_SOURCE=futu_opendcan route genericmarket_priceand technical OHLCV workflows to the local connector.
- New fetch families:
- Registers
futu_opend_localas a private authorized local quote gateway in the source registry and evidence policy. - Documents the Futu private-data boundary in
docs/futu-private-data.md: quote-only, no trade context, no order placement, and raw provider/account artifacts stay underprivate/. - Hardens the Mira data substrate from the post-v0.12.0 fixes:
- bounded retry handling for transient HTTP/body decode failures,
- SEC companyfacts ticker-map resilience,
- source-registry hardening,
- regression tests wired into the local quality gate.
- Fixes issue-triage root causes from #102:
- skill frontmatter validation,
- compact analysis-routing entrypoint with full reference split-out,
- clearer question-lens and evidence-surface boundaries.
Included PRs / commits
- #103 Add Futu OpenD data connector
- #102 Fix Mira issue triage root causes
b030dc1Harden Mira data fetches and source registryac69c00Close superseded Mira data PRs
Verification
python scripts/run_quality_gate.py- py_compile: pass
- mira_data_futu_tests: pass
- mira_data_net_tests: pass
- repo_validation_contract_tests: pass
- repo_validation: pass
- release_gate: pass
python scripts/validate_repo.pyreports 0 errors; existing stale-after warnings are historical research-case freshness warnings.
Known Limits
- Futu OpenD support is optional and requires the user's own local OpenD session plus the optional
futu-apiPython package. - The connector is market-data/read-only research infrastructure. It is not autonomous trading infrastructure and does not create a trading context.
- Broker/provider raw artifacts, entitlement notes, and account-specific data remain private machine state, not tracked product state.
v0.12.0 — route-aware research protocol upgrade
Highlights
- Strengthens quick_map with a route-aware closing contract: every useful quick map now names the next evidence path, calculation depth, readiness change, or route/depth decision.
- Adds behavior eval coverage for valuation, earnings/event, and actionability quick_map prompts, including live/source freshness boundaries.
- Adds marginal-buyer/payoff and left-side actionability protocol upgrades for buy/add/chase style prompts.
- Adds gold residual regime lens and acceleration-to-inflection methodology updates.
- Improves market-date resolution for US, CN, KR, and JP equity sessions.
- Includes the read-only IBKR private data connector work from v0.11.0-era mainline updates.
Verification
- PYTHONPYCACHEPREFIX=/tmp/mira-pycache python3 scripts/score_behavior_eval.py --transcripts evals/transcripts --json
- 35 passed, 0 failed, 8 missing upstream transcripts
- PYTHONPYCACHEPREFIX=/tmp/mira-pycache python3 scripts/test_routing_schema.py
- PYTHONPYCACHEPREFIX=/tmp/mira-pycache python3 scripts/test_repo_validation_contracts.py
- PYTHONPYCACHEPREFIX=/tmp/mira-pycache python3 -m py_compile scripts/score_behavior_eval.py scripts/validate_repo.py scripts/test_routing_schema.py scripts/test_repo_validation_contracts.py
- git diff --check
Known limits
- This is a protocol/workflow release, not an autonomous trading or background monitoring system.
- behavior-eval --require-all is not yet a release gate because 8 upstream behavior cases currently lack recorded transcripts; non-missing cases pass with zero failures.
v0.11.0 — read-only IBKR private data connector
Summary
Mira v0.11.0 adds a read-only Interactive Brokers Gateway/TWS private data connector for broker-authorized research workflows.
Highlights
- Adds a private-first IBKR Gateway adapter for:
- market snapshots
- historical bars
- account summary
- positions
- masked account listing
- private position snapshot export
- Adds CLI fetch families:
ibkr_market_priceibkr_historical_barsibkr_account_summaryibkr_positions
- Adds read-only IBKR utilities:
python -m mira_data ibkr accountspython -m mira_data ibkr position-snapshot
- Registers
ibkr_gateway_localas an authorized local provider with private storage boundaries. - Adds PR-safe IBKR private data documentation and connector registry example.
Data Boundary
- Real account ids, holdings, balances, costs, and broker artifacts remain in
private/or environment variables. - Public templates and docs use placeholders only.
- Account listing is masked by default.
- The connector is read-only and does not place, modify, or cancel orders.
Validation
python -m compileall tools/mira_dataPYTHONUTF8=1 python scripts/validate_repo.py- 0 errors
- 2 pre-existing stale warnings in
cases/us-market-selloff-2026-06-05/
v0.10.0 — critical interaction and actionability protocol upgrade
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/, andlocal/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.pypassed on3e7557e7c8c1fc2b86286efb4d9db20e3a92d1d3.- Required checks passed:
py_compile,repo_validation_contract_tests,repo_validation, andrelease_gate. - Additional PR validation for #87 included
python3 scripts/test_routing_schema.pyand behavior eval report-only with28 passed, 0 failed, 8 missing transcripts.
Known Follow-Up
- Record transcripts for the 8 new behavior eval cases so
just eval-strictcan become the release-grade behavior gate for this interaction layer.
v0.9.0 — market briefing and screening workflow upgrade
Highlights
- Added the market briefing workflow for daily/weekly market reports, close wraps, sector-theme weekly reviews, risk-positioning watches, and escalation queues.
- Added candidate-list screening support in
tools/mira_data, with ajust data-screenentry point and a reusable screening watchlist template. - Strengthened routing behavior around follow-up requirements, portfolio-review task mode, SEC-mode dispatch, and generic screening fallback.
- Replaced the older long-term release pipeline scripts with the parameterized
scripts/validate_release.pyrelease gate.
Validation
python3 scripts/run_quality_gate.pypassed on9571fa6429e33276027c611f30d9fbdccdd5e7c7.- Required checks passed:
py_compile,repo_validation_contract_tests,repo_validation, andrelease_gate.
v0.8.0 — public brand and positioning refresh
Summary
This release packages Mira's public-facing brand and positioning refresh.
What's changed
- Added Mira brand avatar exports for social profiles.
- Added GitHub social preview, README hero, and square social-share assets under
assets/brand/. - Updated public positioning from "investment research workspace" to "AI investment research system" while keeping the internal protocol framing intact.
- Refreshed branded title-card images to use "AI investment research system".
Included PRs
- #76 Add Mira brand avatar assets
- #77 Add Mira repository preview assets
- #78 Update Mira positioning to research system
Validation
python3 scripts/validate_repo.pypassed with 0 errors.- Known warnings remain for stale historical case material in
cases/us-market-selloff-2026-06-05.
v0.7.0 — portable data-acquisition & technical-context substrate
Changes since v0.6.3.
Highlights
Portable data-acquisition & technical-context substrate (#74)
A new stdlib-only layer at tools/mira_data/ that makes Mira's data contract executable — fetch, evidence-tier tag, compute, and emit — on free public data. Lite and supplementary: it strengthens ordinary-user analysis; it does not replace commercial or proprietary data sources.
- Adapters (keyless): SEC companyfacts (
company_financials), Yahoo v8 chart (market_price), BLS (macro_series). - Evidence discipline: every field is born tagged with its registry tier (Yahoo = L5
market_pricing; SEC / BLS = fact), so an aggregator value can't masquerade as an official fact. Mira-computed numbers are ledgered; issuer-disclosed values are reported. - Technical engine: pure-stdlib indicators (MA / returns / relative strength / z-score / realized vol / ATR / drawdown) →
technical-analysis-check.csv+ state tokens; options / short-interest / intraday staysource_gap. - Fundamentals deltas: YoY / CAGR from companyfacts as derived, ledgered records.
- Artifacts: dataset-manifest / ingestion-log / evidence-log (v1.2) / calculation-ledger — all pass
validate_repounchanged. Bundle validator +just data-*recipes. technical-contextoverlay wired into routing; methodology trial (COHR / CRWV / AAPL) recorded — reproducibility milestone met, kept in trial.- Config: SEC fetches are gated on a real contact in
private/mira-data.env(seetemplates/mira-data-config.example).
Quick start:
cp templates/mira-data-config.example private/mira-data.env
just data-fetch company_financials AAPL
just data-technical AAPL
just data-fundamentals AAPL
Also in this release
- Research report interpretation workflow (#75)
- Reasoning-layer ablation trial archived as a pre-registered negative result (#73)
- Minimum entry gate folded into the ChatGPT conversation instructions (#71)
- Stale repository artifacts pruned (#72)
Verification
validate_repo (strict) + routing-schema tests are green on the release commit; the data substrate is live-verified against SEC / Yahoo / BLS endpoints. The core is zero-dependency stdlib.
v0.6.3
What changed
- Add the autoresearch agent loop methodology from #67.
- Add the US market selloff case now present on
main. - Absorb the Mira research validation hardening from #69:
- case-level research package manifests
- evidence posture migration for active logs
- calculation-ledger traceability for derived calculations
- repository validation contract tests
- one-command
scripts/run_quality_gate.py - copyable ChatGPT conversation instructions
Verification
For #69 before merge:
git diff --cached --check
python3 scripts/run_quality_gate.pyFull quality gate passed: py_compile, repository validation contract tests, repository validation, and long-term release checks.
Full Changelog: v0.6.2...v0.6.3