v0.3.1 — first PyPI release (pip install maishac)

pip install maishac      # or: pipx install maishac

Maisha is now on PyPI: https://pypi.org/project/maishac/

Patch over v0.3.0: fixes the release workflow (the permissions: block for Trusted Publishing had dropped contents: read, breaking checkout). This is the first release actually published to PyPI. Feature set is identical to v0.3.0 — see its notes.

Verified: builds clean, and pip install maishac in a fresh environment gives a working maishac CLI (54 tests green).

v0.3.0 — MISRA Compliance:2020 evidence set, proactive authoring, robust SARIF

The open-source compliance-workflow layer for MISRA C / CERT C / BARR-C — free, vendor-neutral, agentic-IDE-native. See How Maisha compares.

Highlights since v0.2.0

MISRA Compliance:2020 evidence set — the three documents an assessor asks for:

• Guideline Compliance Summary (report --format misra-compliance)
• Guideline Enforcement Plan (--format gep) — per-guideline method + live tool inventory (versions/options)
• Guideline Re-categorization Plan (maishac recategorize, --format grp) — MISRA legality enforced (no Mandatory re-cat; Required can't drop to Advisory/Disapplied)

Proactive authoring (write it compliant the first time)

• compliance_guidance / maishac guide + compliance_check_snippet / maishac check
• A 37-concern compliant-pattern library covering all 81 KB rules; AUTHORING_PLAYBOOK.md

Robust SARIF import across real qualified-engine dialects

• Recovers the MISRA/CERT guideline when ruleId is a checker id (e.g. Helix QAC ABV.GENERAL) via relationships → taxonomies; ruleIndex/result.rule; kind/baselineState filtering; defaultConfiguration levels; URI normalization
• Bring-your-own-triage: result.suppressions carried across on import

Open-source ready

• CONTRIBUTING.md, CODE_OF_CONDUCT.md, issue/PR templates
• PyPI packaging + Trusted-Publishing release workflow; pipx install maishac
• Honest COMPARISON.md vs the proprietary tools

54 tests green; wheel builds clean and installs standalone.

v0.2.0 — richer SARIF field mapping

SARIF import/export is no longer lossy:

• Code flows — import parses a qualified engine's codeFlows into an ordered data-flow path (new code_flow column) and surfaces it in the agent fix briefing, so a fixer sees how a defect flows, not just where.
• Rule relationships — export emits cross-standard equivalences as SARIF reportingDescriptor.relationships (e.g. a MISRA rule linked to its CERT equivalent), with every target resolving to a descriptor.
• Round-trip — startColumn + codeFlows survive import → export alongside the existing maishac/v1 partialFingerprints identity.

Benchmarked end-to-end (run_sarif_import_test.py, 8/8 checks; see BENCHMARK-SUITE-REPORT.md §4). 33 unit tests green.