Skip to content

release: v2.1.6 — deepwork resilience, fail-fast anti-mocking, and project tool execution - #449

Merged
ils15 merged 8 commits into
mainfrom
fix/deepwork-resilience-and-tool-cwd
Nov 16, 2026
Merged

ils15 merged 8 commits into
mainfrom
fix/deepwork-resilience-and-tool-cwd

Conversation

@ils15

@ils15 ils15 commented Nov 15, 2026

Copy link
Copy Markdown
Owner

Summary

This PR addresses the core framework issues identified during end-to-end multi-agent orchestration runs:

  1. Deepwork Plan Schema & Multi-Decision Normalization:

    • Supported single-option decision policies and multi-decision blueprints without schema rejection.
    • Resiliently inferred acceptance criteria kinds (test, command, file, manual) from string inputs and preserved raw command text.
    • Normalized agent aliases (e.g. @demeter -> demeter) and choices without losing blueprint intent.
    • Prevented race conditions and concurrent write loss during decision resolutions with atomic locking and safe temp files.
  2. Tool Execution Context & Directory Fallback:

    • Converted static shellTool to dynamic createShellTool using resolveToolCwd, ensuring command executions without explicit directory target run in the active project workspace instead of falling back to $HOME.
    • Flattened JSON schema definition for pantheon_jev_decide to comply strictly with upstream host validators (eliminating top-level oneOf/anyOf restrictions).
  3. Skills & TUI Polish:

    • Improved lexical ranking in pantheon_skill_search with threshold gating to prevent word-internal substring false positives.
    • Deduplicated todo block in TUI preferences to leverage native sidebar tracking cleanly.
  4. Documentation:

    • Updated README.md detailing deepwork resilient normalization, tool schema validation, and shell working directory resolution.

Verification

  • npm run lint (biome ci .): Clean (496 files checked).
  • npm run typecheck (tsc --noEmit): Clean.
  • npm run test:quick: 73/73 passed.
  • node --test tests/pantheon/deepwork-plan.test.ts tests/pantheon/tools-shell.test.ts tests/pantheon/jev-decisions.test.ts tests/pantheon/skill-search-tool.test.ts: 30/30 passed.
  • End-to-end sandbox validation: verified in ~/pantheon-sandbox/ with 303/303 tests passing in 9.8s.

@ils15 ils15 changed the title fix(deepwork,tools): support multi-decision plans, resilient criteria normalization, and project shell execution release: v2.1.6 — deepwork resilience, fail-fast anti-mocking, and project tool execution Nov 16, 2026
@ils15
ils15 force-pushed the fix/deepwork-resilience-and-tool-cwd branch from ca36960 to de352bf Compare November 16, 2026 09:26
@ils15
ils15 merged commit 18536e6 into main Nov 16, 2026
10 checks passed
@ils15
ils15 deleted the fix/deepwork-resilience-and-tool-cwd branch November 16, 2026 09:39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant