Skip to content

v0.6.2a1

Pre-release
Pre-release

Choose a tag to compare

View all tags
@github-actions github-actions released this 16 May 20:42
· 170 commits to main since this release
v0.6.2a1
c8ea590
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Alpha pre-release. pip install aiswmm continues to deliver v0.6.1 per PEP 440.

pip install aiswmm==0.6.2a1   # pin this pre-release
pip install --pre aiswmm      # allow any pre-release

Bug fixes (six)

  • Warm intro fires once per session, not on every greeting (#108). Pre-fix: hi/hello repeated 4× emitted the canned warm-intro 4× without calling the LLM. Source-level regression guard added.
  • First plot_run no longer hangs ~90s on matplotlib/swmmtoolbox cold start (#109/#110). The swmm-plot MCP server preheats those imports at boot.
  • Plot X-axis renders readably across any simulated duration (#112). AutoDateLocator + ConciseDateFormatter replace the 316-label black-blur seen on year-long runs.
  • swmm-end-to-end and 5 sibling pure-orchestration skills now appear in aiswmm skill list (#113). Previously dropped silently from the registry.
  • Compound intent "run X demo and plot" routes correctly (#111). Two-layer fix: keyword-fallback priority repair AND a new intent_disambiguator deep module that fires a forced-enum LLM call only when plot co-occurs with another action verb (5s timeout, fail-soft).
  • aiswmm doctor warns on stale editable installs + mcp.json checkout drift; new aiswmm setup --refresh-mcp flag (#113/#114). Re-aligns mcp.json with the active editable install without touching other config.

Portability hardening

  • Zero hardcoded watershed names in routing/inference code (#118). Case-ID inference, continuation classification, and memory summarization now derive case identity from case_registry.list_cases() (with optional aliases via case_meta.extra). Migration to a new watershed requires only cases/<your-watershed>/case_meta.yaml — no code changes. AST-based regression guard prevents future leaks.

Documentation

Full retrospective with before/after tables, PR map (commit SHAs), migration guide, and reproducibility footprint: docs/releases/v0.6.2-alpha.md.

Reproducibility footprint

  • PyPI: aiswmm==0.6.2a1
  • Git tag: v0.6.2a1
  • Pinned model snapshot: gpt-5.5-2026-04-23 (unchanged from v0.6.1)
  • Docker base ref: AGENTIC_SWMM_REF=v0.6.2a1

What's NOT in this release

  • Architecture audit findings (dispatched as a separate background task after this release; will produce ready-for-agent issues covering README-promise-vs-implementation gaps, deeper portability invariants, and keyword-matching centralization across the 6 files where it's still scattered).
  • Stable channel users on pip install aiswmm continue receiving v0.6.1. The next stable release will incorporate v0.6.2-alpha contents.

Acceptance test (zero OpenAI cost)

Check Result
pytest on 12 PR-relevant test files 53/53 pass in 2.7s
Mocked-LLM end-to-end (AISWMM_OPENAI_MOCK_RESPONSE) Warm intro 1-fire, retro chrome rendered
Compound intent regression "run Tod Creek demo and plot the figue" → mode=prepared_demo (was existing_run_plot)
Plot X-axis tick count (30-day fixture) 6 labels (was 316)
aiswmm setup --refresh-mcp against temp config 11 servers under active repo root, drift WARNs zero
Hardcoded watershed names in agent runtime + memory skill 0 code-level references (AST guard enforced)
aiswmm doctor startup latency 0.06–0.11s
Assets 4
Loading