Chara's Toolkit: 22 portable engineering skills, six optional Codex agents, and offline validation for real software projects
This repository also configures its own agents. See what codex-toolkit tells them →
Skill Codex
Hunt for previously unknown correctness defects by deriving invariants and proving or retiring candidates. Use for proactive bug discovery without a specific known symptom.
Skill Codex
Plan and execute dependency, framework, schema, API, or runtime transitions with compatibility and rollback. Use when a known current state must move safely to a defined target.
Skill Codex
Discover and prioritize evidence-backed improvements when no specific change has been chosen. Use for open-ended requests to improve, modernize, simplify, harden, or clean up a repository.
Skill Codex
Inspect and sanitize provenance or metadata in user-owned text, images, PDFs, and documents. Use for evidence-first removal of invisible Unicode, C2PA, EXIF/XMP, or document metadata.
Skill Codex
Diagnose a concrete failure through reproduction, hypotheses, causal tracing, and falsification. Use when observed behavior is wrong and the cause is uncertain.
Skill Codex
Delegate independent, bounded work to specialized reader and writer subagents with explicit ownership and parent verification. Use when parallel work can be scoped and verified independently.
Skill Codex
Update documentation to match code, API, schema, configuration, deployment, or operational changes. Use when implementation changes create documentation drift.
Skill Codex
Build or audit production Expo/React Native applications. Use when Expo/React Native is already selected and implementation, native integration, updates, or release behavior is in scope.
Skill Codex
Build or audit production Flutter applications. Use when Flutter is already selected and implementation, integration, performance, accessibility, or release behavior is in scope.
Skill Codex
Choose a mobile architecture among Flutter, Expo/React Native, native, and other viable options. Use when platform choice or mobile architecture direction is unresolved.
Skill Codex
Plan safe parallel work across independently ownable resources with dependency ordering and exclusive write ownership. Use when multiple bounded work items can proceed independently.
Skill Codex
Measure and improve a named performance bottleneck while preserving correctness. Use when reducing latency, CPU, memory, I/O, rendering cost, startup time, or infrastructure cost.
Skill Codex
Define product UX and visual direction from user goals, brand evidence, states, responsiveness, and accessibility. Use when the main question is what the experience should communicate or feel like.
Skill Codex
Build or audit production web interfaces across architecture, data, interaction, accessibility, performance, security, and browser behavior. Use when web implementation quality is the main decision.
Skill Codex
Map repository architecture, ownership, dependencies, change impact, and edit conflicts. Use when the task needs evidence about what exists or what a proposed change can affect.
Skill Codex
Review a defined code area for actionable defects and perform behavior-preserving refactors. Use for bounded code review, maintainability cleanup, decomposition, or refactoring.
Skill Codex
Reconstruct a maintainable interface from screenshots or visual references. Use when fidelity to supplied visual evidence is the primary requirement.
Skill Codex
Resolve local shell, runtime, package-manager, browser, quoting, or encoding blockers before retrying the real task. Use when the execution path itself is unknown or broken.
Skill Codex
Strengthen TypeScript/JavaScript type safety and deterministic lint discipline. Use for systematic type-safety cleanup, unsafe escape-hatch reduction, or anti-slop rule adoption.
Skill Codex
Apply auditable completion discipline to substantial already-scoped work. Use when the user wants every requested deliverable finished, measurable completion gates, or final claims rechecked.
Skill Codex
Design risk-based verification and judge release readiness from traceable evidence. Use when deciding what must be proven or whether an integrated change can ship.
Skill Codex
Review code for exploitable security flaws at trust boundaries. Use when assessing authentication, authorization, untrusted input, secrets, file/network handling, or security-sensitive diffs.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: