Skip to content

Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS

4.3 KB
Raw Blame

Scheduled audit names, labels, and order

All tasks in AUDIT_DEFINITIONS belong to the better-* display-name family. On install/update, schedule responses derive these names from the catalog. Existing task IDs remain stable: saved prompts, provider pins, cadence, execution history, run-after references, app overrides, API calls, and bookmarked task IDs continue to work. No task is newly enabled or automatically run.

The newly prefixed display names are: better-security, better-code-quality, better-test-coverage, better-performance, better-accessibility, better-documentation, better-ui-bugs, better-mobile-responsive, better-error-handling, better-typing, better-console-errors, better-ux, better-data-safety, better-simplify, better-module-hygiene, better-api-contract, better-ui-lifecycle, better-observability, and better-copy, plus the service and data-platform lenses better-infrastructure, better-data-integrity, better-reliability, better-privacy, and better-cost-efficiency, and the gap lenses better-feature-completeness and better-workflow-recovery. The existing better-* names, including better-dev-environment, remain unchanged.

Operational tasks (claims, review pipelines, reconciliation, releases, repo sync), planning/ideation, private security assessment, JIRA, and media generation retain their names: they have distinct execution or security contracts and are not members of the configurable audit catalog. Independent code-reviewer A/B tasks also retain their identities; they are general review workflows.

Shipped labels include codebase-improvement, slashdo, and the audit's mapped slashdo lenses. The slashdo label means alignment with that library's audit family, not that dispatch necessarily invokes a slash command. Custom labels are optional schedule metadata, saved in the existing task config; absent means empty, and clearing uses an empty array. This additive field requires no layout conversion or seed rewrite. Shipped labels and names are derived in memory, so older schedules gain them without overwriting any customization. Labels do not change dispatch, issue labels, eligibility, or failure backoff.

The label filter is linkable with ?label=slashdo. It intersects the status filter and text search; search accepts old IDs, display names, descriptions, and both shipped and custom labels. Shipped labels remain available even when custom labels are cleared.

Recommended order for one code area

The bundled lib/slashdo/lib/better-audit.md runs the test audit after the other selected audit scopes so it can consume their findings; those other audit scopes are independent. Follow that for an issues-only audit batch. The sequence below concerns implementing changes in the same area: tests still need to protect behavior before a refactor begins.

This is engineering guidance, not a dependency graph or a claim that every repo needs every pass. Existing evidence and urgent defects take priority.

  1. Broad code-quality triage if needed; ensure tests detect behavior and add missing boundary coverage. Fix security, data-loss, and runtime defects first.
  2. Structural drift: consolidate sources of truth before polishing code that may disappear. Simplify: remove dead and duplicate code next.
  3. Module hygiene: establish ownership, reuse, and boundaries. Remove unnecessary dependencies once the surviving owners are clear.
  4. Complexity: remeasure branching in surviving functions and simplify the costly ones. Cognitive load: review readability of the final shape.
  5. Revalidate runtime behavior and meaningful tests; measure performance where there is a demonstrated bottleneck. Refresh documentation after changes land.

Thus module hygiene generally precedes function-level cyclomatic complexity, and structural drift generally precedes both. A localized hot function can be improved independently when it has no ownership or source-of-truth problem. API, React, accessibility, UI, copy, and observability passes follow the needs of the changed area rather than a universal rank. Work on disjoint areas may run independently; serialize overlapping edits, merge, and reassess before the next pass. A completed audit that only files issues has not completed remediation. The page repeats the relevant recommendation alongside each audit description; custom bylines remain intact and existing run-after settings are unchanged.