SDD Observatory
Stars
1.5K
Contributors
3
Open issues
21
Age
4 months
Last push
3 days ago
Latest release
Week of Aug 31, 2025: 0 commits Week of Sep 7, 2025: 0 commits Week of Sep 14, 2025: 0 commits Week of Sep 21, 2025: 0 commits Week of Sep 28, 2025: 0 commits Week of Oct 5, 2025: 0 commits Week of Oct 12, 2025: 0 commits Week of Oct 19, 2025: 0 commits Week of Oct 26, 2025: 0 commits Week of Nov 2, 2025: 0 commits Week of Nov 9, 2025: 0 commits Week of Nov 16, 2025: 0 commits Week of Nov 23, 2025: 0 commits Week of Nov 30, 2025: 0 commits Week of Dec 7, 2025: 0 commits Week of Dec 14, 2025: 0 commits Week of Dec 21, 2025: 0 commits Week of Dec 28, 2025: 0 commits Week of Jan 4, 2026: 0 commits Week of Jan 11, 2026: 0 commits Week of Jan 18, 2026: 0 commits Week of Jan 25, 2026: 0 commits Week of Feb 1, 2026: 0 commits Week of Feb 8, 2026: 0 commits Week of Feb 15, 2026: 0 commits Week of Feb 22, 2026: 0 commits Week of Mar 1, 2026: 0 commits Week of Mar 8, 2026: 0 commits Week of Mar 15, 2026: 0 commits Week of Mar 22, 2026: 0 commits Week of Mar 29, 2026: 0 commits Week of Apr 5, 2026: 0 commits Week of Apr 12, 2026: 0 commits Week of Apr 19, 2026: 0 commits Week of Apr 26, 2026: 50 commits Week of May 3, 2026: 22 commits Week of May 10, 2026: 3 commits Week of May 17, 2026: 8 commits Week of May 24, 2026: 0 commits Week of May 31, 2026: 0 commits Week of Jun 7, 2026: 0 commits Week of Jun 14, 2026: 2 commits Week of Jun 21, 2026: 0 commits Week of Jun 28, 2026: 3 commits Week of Jul 5, 2026: 0 commits Week of Jul 12, 2026: 6 commits Week of Jul 19, 2026: 3 commits Week of Jul 26, 2026: 11 commits Week of Aug 2, 2026: 1 commit Week of Aug 9, 2026: 0 commits Week of Aug 16, 2026: 0 commits Week of Aug 23, 2026: 0 commits
Commits per week, last 12 months · 109 total
View weekly commit data
Week startingCommits
Aug 23, 20260
Aug 16, 20260
Aug 9, 20260
Aug 2, 20261
Jul 26, 202611
Jul 19, 20263
Jul 12, 20266
Jul 5, 20260
Jun 28, 20263
Jun 21, 20260
Jun 14, 20262
Jun 7, 20260
May 31, 20260
May 24, 20260
May 17, 20268
May 10, 20263
May 3, 202622
Apr 26, 202650
Apr 19, 20260
Apr 12, 20260
Apr 5, 20260
Mar 29, 20260
Mar 22, 20260
Mar 15, 20260
Mar 8, 20260
Mar 1, 20260
Feb 22, 20260
Feb 15, 20260
Feb 8, 20260
Feb 1, 20260
Jan 25, 20260
Jan 18, 20260
Jan 11, 20260
Jan 4, 20260
Dec 28, 20250
Dec 21, 20250
Dec 14, 20250
Dec 7, 20250
Nov 30, 20250
Nov 23, 20250
Nov 16, 20250
Nov 9, 20250
Nov 2, 20250
Oct 26, 20250
Oct 19, 20250
Oct 12, 20250
Oct 5, 20250
Sep 28, 20250
Sep 21, 20250
Sep 14, 20250
Sep 7, 20250
Aug 31, 20250

Metrics updated Aug 27, 2026 · collected automatically from the GitHub API

Core approach

Coordinates specialist agents through a five-phase discovery pipeline that inventories a legacy system, excavates its code, infers business and architectural knowledge, generates operational specifications, and reviews contradictions and gaps with the user. Checkpoints live in .reversa/, while specs, confidence reports, and code-to-spec and impact matrices are written under _reversa_sdd/ by default; follow-on workflows use those artifacts to drive forward development and migration.

Workflow

  1. npx reversa install — detect supported agents, install project-local skills and steering files, and initialize .reversa/ state
  2. Reconnaissance — the Scout inventories the repository, dependencies, languages, frameworks, and entry points
  3. Excavation — the Archaeologist analyzes one module at a time, recording algorithms, control flow, data structures, and a data dictionary
  4. Interpretation — the Detective recovers business rules and retroactive decisions while the Architect produces architecture, data, integration, and debt models
  5. Generation — the Writer turns the recovered knowledge into confidence-marked component specifications, API and user-story artifacts, and traceability matrices under the default _reversa_sdd/ path
  6. Review — the Reviewer challenges contradictions, reclassifies unsupported claims, asks the user about gaps, and emits a confidence report
  7. /reversa-forward — evolve one feature through requirements, clarification, quality review, a delta plan, atomic actions, cross-checking, coding, and optional /reversa-sync convergence

Supported tools

Strengths

  • Makes brownfield specification recovery the primary workflow rather than a preparatory option inside a greenfield-oriented method
  • Combines code inventory, business-rule extraction, architecture reconstruction, confidence labels, unresolved-gap reporting, and human validation
  • Explicit code-to-spec and spec-impact matrices give subsequent agents a traceability model for change analysis
  • Follow-on workflows cover forward feature delivery, migration, defect handling, documentation, and behavior-preserving refactoring without discarding the recovered system model
  • Project-local checkpoints and restricted output directories make long analyses resumable and limit where discovery agents are instructed to write

Limitations

  • A very young project with concentrated maintenance and limited independent production evidence
  • Its linked migration paper is maintainer-authored, so it explains the method but does not independently validate its results
  • The large and quickly expanding roster of specialist teams and commands creates substantial selection and review overhead
  • Claude Code is documented as the most-tested engine; many other integrations are steering-file and shared-skill compatibility surfaces rather than equally demonstrated native runtimes
  • Output can be redirected or excluded from version control; public discovery sees only teams that commit the default _reversa_sdd/ path

Notes

Reversa fills a gap in the current catalog: it begins with code that already exists but lacks trustworthy specifications. Its discovery pipeline separates observation from interpretation before a writer produces operational contracts, then marks claims as confirmed, inferred, or gaps for a reviewer and the user to resolve. After that recovery pass, the same repository-local artifacts can drive feature work, migration, debugging, or refactoring. AI Unified Process also offers reverse engineering, but Reversa’s dedicated confidence model and code-to-spec and spec-impact matrices make legacy traceability the center of the method. The repository is active and publicly usable today, yet its concentrated maintainer base and lack of independent outcome evidence make the emerging label deliberately provisional. This assessment observes the repository at 4f57467.

Added Aug 1, 2026 · Assessment last reviewed Aug 1, 2026 · How we track