harnessy

Autonomous code refactoring agent that rewrites messy functions, splits large files, removes duplication, improves naming, enforces structure, and preserves behavior.

Interact with Anytype's local REST API to manage spaces, objects, types, properties, tags, lists, and search.

Generate API integration test suites from Harnessy regression artifacts using a QA profile plus optional delivery-profile adapters.

Generate browser integration suites from Harnessy regression artifacts using a QA profile plus optional delivery-profile adapters for fixtures, routes, and helper imports.

End-to-end product development orchestrator with human-in-the-loop reviews.

Safe repo pull/push workflow that targets the repository integration branch, uses deterministic planning, and keeps PRs moving with a bounded autonomous resolution loop.

Plan, verify, and explicitly install skill and script runtime dependencies from Harnessy manifests.

Life orchestration — monthly reviews, weekly plans, daily focus briefs from priorities.md and project state

Deterministic QA runtime for parsing regression specs, scanning tests, detecting drift, and generating coverage from a repo-local QA profile.

Full-cycle QA orchestrator for Harnessy-compatible repositories: discovers feature surfaces, uses Playwright/browser walkthroughs when available, maps scenario gaps, updates regression artifacts, runs drift/validation, and reports coverage.

Plan, package, verify, deploy, inspect, and roll back services using Harnessy CI/deploy profiles with Hostinger as the first provider adapter.

Capture ad-hoc feedback about any skill for the decision trace system.

Validate generated or hand-written tests for coverage completeness, correctness, and false-green risks using the canonical QA runtime, regression artifacts, and optional delivery-profile rules.

Launch Claude, OpenCode, or Codex in a named tmux session from the command line or agent workflows.

Review articles, posts, and publishing drafts against supplied research, facts, audience needs, editorial quality, and a caller-provided strategy profile. Use when content needs fact-checking, anti-robotic-language cleanup, coherence review, accessible explanations for general audiences, or alignment with project/campaign context provided at runtime.

Generate comprehensive UX/design specifications with Mermaid user flow diagrams, component specs, interaction patterns, and accessibility requirements from product specs.

Provider-agnostic goal orchestrator — decomposes goals, drives Codex/Claude/OpenCode phase work in a loop, verifies completion objectively, and persists run policy/learning state.

Use when a GitHub issue should be driven through clarification recovery, spec-build-test-QA-acceptance, and enforced human review gates with quality gates.

Bridge to Jarvis CLI for scheduling, journaling, reading list orchestration, and context management.

Distill technical_spec.md into a focused MVP with prioritized work items.

Transform brainstorm.md into a comprehensive Product Specification Document.

Multi-perspective quality review for product_spec.md files.

Seven-lens engineering review for technical_spec.md files, including an explicit simplicity and architectural fitness gate.

Generate production-ready technical specifications from product specs.

Use when the user wants to manage their personal knowledge base: ingest a URL or file, compile wiki articles, ask questions about a domain, run lint, or open in Obsidian. Trigger on: "add to wiki", "compile wiki", "what does my wiki say", "lint wiki".

Maintain a semantic QA feature catalog for Harnessy-compatible repositories by consolidating raw spec suites into stable feature slugs, prefixes, result files, and optional result-sink tabs.

Adversarial QA audit that turns security findings for a feature, module, or diff into canonical Layer:security regression scenarios and archived findings.

Autonomous security auditor that identifies vulnerabilities (OWASP, CVEs) and misconfigurations.

Autoresearch orchestrator — discovers skills with autoresearch.enabled, runs the experiment loop, and uses a multiplicative composite metric with hard constraint gates.

Publish a skill after validation and approvals; updates the catalog and logs the publish event.

Fixture skill used by CI to verify project-local Harnessy skill registration.

Scaffold a new skill in the monorepo with a valid manifest and catalog entry. Use when creating any new skill.

<short description>

Analyze decision traces and propose skill mutations based on accumulated feedback patterns.

Promote trace-driven improvements from the installed skill copy back to the shared Harnessy source.

Generate browser and API regression scenarios from approved product and technical specs using the Harnessy delivery profile and regression artifact contract.

Autonomous research orchestrator for Jarvis Wiki domains. Spawns a Claude agent with web tools to find new sources, drops them into raw/, runs the compile pipeline, and emits autoresearch traces for the autoflow ratchet.

Collaborative brainstorming facilitator that develops raw ideas into well-defined concepts.

Playwright-based browser QA skill for guided setup, auth handoff, scripted runs, and artifact capture.

Iteratively diagnose and fix failing GitHub Actions runs until CI passes or a safe escalation point is reached. Use when a branch has failing CI and an automated fix loop is acceptable.