Totem Shield

Version updated for https://github.com/mmnto-ai/totem to version @mmnto/cli@1.14.14.

  • This action is used across all versions by 0 repositories.

Action Type

This is a Composite action.

Go to the GitHub Marketplace to find the latest changes.

Action Summary

Totem is a CLI tool that provides a persistent memory and enforcement layer for AI coding agents, addressing the issue of repeated mistakes and degraded architectural integrity in AI-assisted development. It automates the creation and enforcement of coding rules by converting plain-English lessons into deterministic, offline linting rules, ensuring that mistakes identified in code reviews or production are not repeated. This enables a more efficient and reliable collaboration with AI agents while maintaining code consistency and quality.

What’s Changed

Patch Changes

  • e073dc0: Flip Pipeline 5 auto-capture on totem review from opt-out to opt-in.

    --no-auto-capture is renamed to --auto-capture; the default is now OFF. Observation rules captured from review findings are context-less (regex drawn from the flagged line, message taken from the reviewer, fileGlobs scoped to the whole codebase) and routinely pollute compiled-rules.json with rules that fire on unrelated files. The Liquid City Session 6 audit measured an 8-rule wave across 5 review invocations producing 13 new warnings on the next totem lint, up from 0.

    To preserve the old behavior, pass --auto-capture explicitly. Auto-capture will resume as a default once ADR-091 Stage 2 Classifier + Stage 4 Codebase Verifier ship in 1.16.0 and the LLM-emitted rule loop has gates that prevent context-less emissions.

    Closes #1579.

  • Updated dependencies [e073dc0]

    • @mmnto/totem@1.14.14