Infrastructure and contracts that make agent workflows bounded, inspectable, and reusable.

• Agent Eval Runtime — Runtime for executing agent evaluation runs, paired comparisons, and retained result artifacts.

• Agent Router — Local-first, evidence-backed multi-agent router that plans work, scores providers, dispatches bounded execution, and returns receipts.

• AI Context Runtime — Local-first Python and SQLite runtime for context events, knowledge inputs, packets, handoffs, and report-only session intelligence.

• AI Workflow Leverage — Evidence-bounded discovery, testing, quantification, and review-gated promotion-candidate pipeline for Jakye's portable agentic setup.

• Context Compiler Contract — Portable ESM validators for compiled-context results and routing manifests

• Jakye Skill Family — Local code-and-contract family of audit-derived preference skills for bounded conversation, recommendations, drafts, and private reflection.

• Portable Agentic Workbench — Vendor-neutral context, routing, safety, evaluation, and durable-handoff contracts

Developer-facing checks and review surfaces that make software quality easier to act on.

• Behavior Coverage Atlas — Repository-owned map from named product behavior to tests and recent execution evidence.

• Quality Lens — IDE-native, human-complete inspection and action surface for deterministic Quality Runner findings.

• Quality Setup — Human-first setup and diagnostics workflow for making repository quality checks understandable and runnable.

• Readiness Inspector — Local inspector that explains whether a repository or change is ready for a stated goal.

• Remediation Canvas — IDE-native workspace for turning a bounded set of quality findings into reviewable human action.

• Review Attention Map — Human-first review surface that helps a reviewer spend attention where evidence says it matters.

• Review Sandbox — Deterministic, disposable local workflow for exercising one meaningful change scenario without an agent or cloud preview.

• Rule Lab — Workbench for understanding, editing, and verifying developer-quality rules without an AI agent.

• Test Intent Registry — CLI that connects repository-owned test intent to explainable lifecycle evidence.

• Workflow Gateboard — Local board for running and understanding repository-declared development gates.

Tools that explain changes, preserve evidence, and reduce integration uncertainty.

• Change Integration Simulator — Disposable local simulation of integrating one change into one explicit target, reporting conflicts and declared gate evidence.

• Change Radius — Change Radar that explains the bounded behavioral and integration radius of a code change with versioned evidence and explicit uncertainty.

• Contract Watch — Local change watcher for declared interfaces, schemas, and compatibility boundaries.

• Debug Trail — Investigation workbench that preserves hypotheses, experiments, evidence, and next steps for human debugging.

• Deletion Proof Workbench — Local workbench for forming and testing bounded code-deletion claims while keeping unknown consumers and recovery visible.

• Evidence Replay — Local viewer for inspecting, navigating, and selectively rerunning the evidence behind a development claim.

• Failure Capsule — Local packaging workflow that preserves the minimum useful evidence for a failed or inconclusive development action.

• Pattern Atlas — Standalone, agent-independent map of structural implementation families and the evidence behind their organizational meaning.

• Steward — Policy governance core that resolves versioned policy-as-code, evaluates deterministic TypeScript and JavaScript dependency boundaries, and preserves evidence-backed receipts.

Guarded automation and local control planes with explicit human boundaries.

• Attentiond — Local macOS attention utility for coding agents.

• Automation Flight Recorder — Local recorder and viewer for bounded development automation runs, including steps, inputs, outputs, timings, and failures.

• Browser Control — Local Chrome MV3 provider that lets a Codex client observe authorized browser state, propose bounded mutations, obtain human approval, apply once, and verify the resulting Chrome state.

• CI Incident Router — Deterministic bridge from failed GitHub Actions runs to bounded Codex prompts.

• Mac Control — macctl, a command-first owner-only macOS control plane with guarded local execution, durable redacted evidence, and explicit provider boundaries.

Sports products and analytical systems spanning training, scouting, and modeling.

• Cap Fit Builder — Team-context roster construction engine for evaluating player fit against needs, role value, contracts, and acquisition realism.

• CLFE — Audit-first public-data context-intelligence package for explaining how player value changes across lineup contexts.

• Fantasy — Dynasty fantasy football intelligence product for format-aware team decisions.

• LIS — Python analytics package for NBA game-control scoring with live-data ingestion, cached rescoring, confidence tiers, and player-level diagnostics.

• Women's Stats — Women’s basketball analytics scaffold supporting normalized data, multiple analysis modes, reports, and explicit uncertainty.

Operational software for teams, professionals, and repeatable business workflows.

• Career Ops — Human-in-the-loop job search command center that turns official postings into evidence-bound fit evaluations, tailored materials, tracker records, and draft-only outreach.

• CrimClock — Legal-time intelligence for procedural deadlines, sentencing, parole eligibility, custody impact, and case timelines.

• Participant Dedup — Google Sheets review tool that finds likely duplicate participant rows, explains the match, and applies reviewer-approved deletions atomically on a copy with an audit trail.

Consumer experiences for music, taste, and media discovery.

• Soundscape — Music-taste platform for capturing reactions, comparing trusted perspectives, discovering music, and revisiting how taste changes over time.

Editorial and portfolio systems for presenting authored work.

• Dispatches from Cyberspace — Versioned publishing platform for authored dispatches with drafts, publication, comments, reactions, JSON Feed, sitemap, and authenticated authorization.

Research pipelines and inspectable analytical applications.

• LaxDS — Lacrosse analytics research and data-science scaffold with explicit provenance and data-governance concerns.

• Marketing Autoresearch — Experimental, portable evidence-bounded weekly SEO and AEO research loop for selected portfolio and product surfaces.

• RTE Transferable Signals — Capstone layer for translating role signals into draft-slot-relative upside, floor, and risk labels for a versioned board.

• Signal Lab — Basketball analytics project that estimates metric reliability across sample size, player role, competition, and context, with a bounded role-translation MVP and reporting surface.