v1.5.1

1.5.1 (2026-06-26)

OpenBB Workspace addon modernization plus API pipeline performance and EDGAR improvements.

OpenBB Workspace

• Copilot prompts and mcp_servers for Disclosure Alpha Analyst at /mcp
• Widget mcp_tool citations linking to score_company_filing_tool
• HTTP MCP mounted on disclosure-alpha-api when [api,mcp] installed

API and pipeline

• In-process TTL cache on metrics_filing_ticker
• Pipeline timing logs (PIPELINE_TIMING)
• Panel screener concurrency (PANEL_MAX_WORKERS)
• Metadata-first EDGAR filing resolution

Install

pip install -U "disclosure-alpha[api,mcp]==1.5.1"

Docs: https://disclosure-alpha.readthedocs.io/en/latest/guides/openbb/index.html

1.5.0

Highlights

• Ship text_metrics_v4 with boilerplate_cross_firm_ratio and boilerplate_combined_ratio.
• Refresh public evidence docs to the latest L2 boilerplate construct result (Spearman ρ≈0.96, n=478).
• Include extraction/scoring response parity updates (confidence_details, section_diffs_v2, and form-aware rescoring paths).
• Bump package/docs pins to 1.5.0 and regenerate HTTP/docs artifacts.

Notes

• Post-filing volatility association remains unchanged at Q5/Q1≈1.15 (n=435); L3 cache report remains separate scope.

1.4.0

1.4.0

Python SDK configuration for tunable scoring and reproducibility metadata.

Highlights

• PipelineConfig / ScoringConfig on Python pipeline entry points — tune component weights, flag scoring constants, and v2 calibration context
• versions.analytics_config_id in pipeline, MCP taxonomy, and panel outputs (builtin_default when unset)
• Docs and examples updated; full-coverage fixture includes cyber incident language

Default scores unchanged when no custom config is passed.

Install: pip install "disclosure-alpha==1.4.0"

v1.3.0

1.3.0

Package release consolidating artifact bumps and default-surface updates.

What shipped

Area	Change
Dictionaries / metrics	built_in_dictionaries_v3 and text_metrics_v3 — dictionary package split, flag suppressions, legal phrases, modal tiers, topic tuning
Scoring default	deterministic_scoring_v2 is now the default on CLI, HTTP, MCP, and score_filing_html() / score_for_model(); legacy deterministic_scoring_v1 remains opt-in
Validation data	Reports and baselines removed from the public repo (see INTERNAL_VALIDATION.md on the internal branch)
Docs	Evidence, scope, HTTP/MCP guides, versioning pins, and glossary aligned to current artifact versions
Release tooling	Version sync test, HTTP endpoints doc drift check, hardened PyPI publish workflow

Current artifact defaults

Artifact	Version
Parser	section_extractor_v1
Metrics engine	text_metrics_v3
Dictionary	built_in_dictionaries_v3
Scoring (default)	deterministic_scoring_v2
Scoring (legacy)	deterministic_scoring_v1

Breaking behavior note: scores may differ from 1.2.0 for callers who did not pin scoring_model_version=deterministic_scoring_v1. Pin package and scoring versions for reproducibility.

• Docs: https://disclosure-alpha.readthedocs.io/en/latest/
• PyPI: https://pypi.org/project/disclosure-alpha/1.3.0/

v1.2.0

1.2.0 — Evidence-gated v2 validation

• FY2025 v2 committed validation: L2 construct pairs (n=428), matrix gates on partial EDGAR matrix corpus (n=330), L3 volatility Q5/Q1 ≈ 1.05 (n=435, corpus mode).
• v1 remains default; v2 opt-in on HTTP/MCP unchanged.
• EDGAR full-matrix corpus builder (from 1.1.x line).
• Docs: Evidence & limitations

Caveats: matrix corpus incomplete vs full S&P 500; EDGAR E1 fetch gate still fails; v2 score levels are not comparable to v1.

v1.1.0

Breaking changes

• Removed view from GET /v1/company/{ticker}/disclosure-matrix and POST /v1/panel/disclosure-matrix (request and response). Scoring is deterministic only.

Fixes

• disclosure_quality_score is correct when boilerplate_risk_score is 0.0.
• Unified confidence_score via score_deterministic (removed unused llm_confidences hook).

Documentation

• Public docs aligned with deterministic scoring surface; score catalog renamed to Score catalog.
• Removed composite/OSS product-scope notes from user-facing pages.

Install

pip install "disclosure-alpha==1.1.0"
pip install "disclosure-alpha[api,mcp]==1.1.0"  # HTTP + MCP

Full changelog: https://github.com/alwank/disclosure-alpha/blob/main/docs/appendix/changelog.md

v1.0.0

Initial open-source release of Disclosure Alpha.

Highlights

• Deterministic SEC filing analytics for 10-K, 10-Q, and 8-K HTML
• CLI, Python SDK, HTTP API, and MCP (Analyst + Builder) surfaces
• Dictionary v2 / text metrics v2 with enriched built-in word lists
• Validated on ~425 S&P 500 FY2025 10-K Item 1A extractions

Install

pip install "disclosure-alpha==1.0.0[dev]"
pip install "disclosure-alpha==1.0.0[api,mcp,dev]"  # HTTP + MCP

Version artifacts

Artifact	Version
Package	1.0.0
Dictionary	built_in_dictionaries_v2
Metrics engine	text_metrics_v2
Scoring model	deterministic_scoring_v1

Full changelog: https://github.com/alwank/disclosure-alpha/blob/main/docs/appendix/changelog.md