Releases: tomyimkc/ledgerlens
Release list
LedgerLens v0.2.0 — Autonomous Data Incident Commander
LedgerLens v0.2.0 — Autonomous Data Incident Commander
This release upgrades LedgerLens from failure-ledger triage to an end-to-end, verifier-gated data incident command workflow:
- DataHub-grounded incident context and lineage blast radius
- planner plus two verifier variants
- deterministic fail-closed policy authorization
- HMAC-bound, idempotent GitHub, Slack, PagerDuty, and Jira adapters
- controlled DataHub MCP write-back with audit and rollback metadata
- receipt-bearing next-agent memory
- 120-asset, four-domain catalog and 24 incident scenarios
- DataHub-context ON/OFF benchmark
- judge-ready Incident Commander dashboard, CLI, Docker image, and deployment package
Published evidence includes a live 020s verification receipt, a real GitHub issue creation/closure receipt, and a local DataHub OSS v1.6.0 catalog/write-back/retrieval receipt.
Claim boundary: candidateOnly: true, canClaimAGI: false. Distinct model variants do not establish provider-family independence. Metadata and action receipts do not prove incident causality, user impact, recovery, production readiness, validated uplift, or AGI.
LedgerLens v0.1.0 — DataHub Agent Hackathon Prototype
LedgerLens v0.1.0 — DataHub Agent Hackathon Prototype
LedgerLens turns structured failure records into an evidence-grounded remediation
queue using DataHub OSS, the official read-only DataHub MCP server, and a
fail-closed deterministic core.
Release evidence
- Public Apache-2.0 repository.
- Release target: merged public
mainata6e253c. - GitHub Actions passed on Python 3.11, Python 3.12, and the non-root container build.
- Post-merge clean-clone gate: 127 tests passed, plus Ruff, package, secret, and
public-package checks. - Final demo: 41.012 seconds, 1920x1080 H.264, AAC English narration, burned
English captions, and a selectable English caption track. - Generated concept footage: 5 of 41 seconds (12.20%), visibly labeled. All product
functionality claims use real LedgerLens and DataHub captures.
Claim boundary
LedgerLens is a working prototype. Its benchmark receipts are candidate evidence,
not independent validation of Sophia or of any source finding.
candidateOnly: true
canClaimAGI: falseThe demo video still requires upload to the contest-accepted public video platform
and an incognito playback check before Devpost submission.
See the attached OWNER-SUBMISSION-HANDOFF.md for the exact remaining account-only
steps, paths, hashes, and submission fields.