Indium-AI-Labs

Indium-AI-Labs/indium-agentkit

Agent Claude Code

Reviews a diff, branch, or pull request for correctness, security, error handling, performance, and test coverage, verifying each finding before reporting it. Use proactively after code changes and before merging. Read-only.

Indium-AI-Labs/indium-agentkit

Agent Claude Code

Drafts an operational runbook for one service and failure mode from repository evidence: detection, diagnosis, mitigation, recovery, escalation, and rollback limits, with every step labeled by verification status. Use when a service, alert, or procedure lacks a runbook or an existing one no longer matches the code.…

Indium-AI-Labs/indium-agentkit

Agent Claude Code

Reviews a bounded change, endpoint, or integration for exploitable security flaws in authentication, authorization, input handling, secrets, tenant isolation, and LLM or agent features such as prompt injection and over-broad tool permissions. Use proactively before merging security-sensitive changes or to confirm a…

Indium-AI-Labs/indium-agentkit

Agent Claude Code

Designs referentially intact, PII-free, reproducible synthetic data plans from a schema: insertion order over the foreign-key graph with cycle handling, volumes and skewed distributions, constraint-safe field rules, and seeding determinism. Use when seed, fixture, or load-test data is needed, or when an existing seed…

Indium-AI-Labs/indium-agentkit

Agent Claude Code

Verifies completed work against its acceptance criteria by running each of the project's declared lint, type-check, build, and test commands separately and tracing every criterion to evidence. Use after implementation and before reporting a task done, committing, or merging. Read-only.

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Audit a web interface against WCAG 2.2 Level AA with automated axe-core scans plus manual keyboard and screen reader passes, mapping every finding to a success criterion with evidence and producing an accessibility-audit-report. Use when asked to audit accessibility, check WCAG, ADA, or Section 508 readiness, or…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Design an API contract (resources, operations, request and response schemas, errors, authorization, pagination, idempotency, and versioning) from requirements before implementation, and produce an api-contract handoff. Use when a feature needs new or changed endpoints, an OpenAPI document, or a compatibility decision.…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Write, test, and run syntax-aware codemods (ast-grep, comby, jscodeshift, ts-morph, libcst, rust-analyzer) for large mechanical migrations such as API renames, import moves, and call-signature changes, with fixture tests, idempotency, and git diff as the review. Use when the same structural edit must land across many…

Indium-AI-Labs/indium-agentkit

Skill Codex

Create or revise indium-agentkit skills, subagents, templates, and their validation and catalog files to the repository authoring standard. Use when adding, rewriting, or reviewing content in this agentkit repository. Internal to this repository; not for consumer projects.

Indium-AI-Labs/indium-agentkit

Skill Codex

Implement or change server-side endpoints, handlers, and services against an agreed API contract, with boundary validation, server-side authorization, transactional writes, idempotent retries, and tests at the HTTP seam, then produce a verification report. Use when implementing or changing server endpoints. Not for…

Indium-AI-Labs/indium-agentkit

Skill Codex

Design or harden CI workflows, GitHub Actions by default, with fast merge-blocking checks, a correct required-checks aggregator, least-privilege tokens, SHA-pinned actions, safe caching, fork-safe triggers, timeouts, and concurrency. Use when creating or changing CI workflows, required checks, or branch protection…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Assess code, configuration, and data flows for readiness against GDPR, SOC 2, HIPAA, or PCI DSS controls, covering the regulated-data inventory, PII in logs, encryption, access control, audit logging, and retention and erasure, with a findings register that cites control references. Use when preparing for an audit, a…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Write consumer-driven contract tests with Pact (consumer tests, provider verification with state handlers, broker publishing, and can-i-deploy gates) and keep frontend mocks aligned, so interface breaks between services or between frontend and backend fail in CI instead of production. Use when adding or changing an…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Implement or review application cryptography with vetted high-level libraries, covering authenticated encryption with key IDs for rotation, nonce strategy, signatures, key derivation, password hashing, TLS and mTLS configuration, and hybrid post-quantum key exchange. Use when code must encrypt, sign, derive keys, hash…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Design batch or incremental ETL and ELT pipelines and warehouse models: grain, partitioning, idempotent loads, late-arriving data, data quality tests, PII controls, and cost monitoring, before building them. Use when adding or changing a dbt model, warehouse table, scheduled transformation, or ingestion job. Not for…

Indium-AI-Labs/indium-agentkit

Skill Codex

Design relational schemas, constraints, and indexes from real access patterns, plus a staged migration plan with reconciliation queries, defaulting to PostgreSQL. Use when adding or changing tables, columns, keys, relationships, or indexes, or when a feature needs a new data model. Not for executing a rollout (use the…

Indium-AI-Labs/indium-agentkit

Skill Codex

Decide whether to hand parts of a task to subagents, then write self-contained briefs, run independent pieces in parallel, and verify and merge the results before relying on them. Use when a task has independent, parallelizable, or context-heavy parts such as broad code searches, multi-area reviews, or separate…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Audit a project's dependencies for known vulnerabilities, license risk, staleness, unused packages, and version-constraint problems using lockfiles and read-only scanners such as npm audit, pip-audit, osv-scanner, govulncheck, and cargo audit, then write a dependency-audit report. Use when asked to check dependencies…

Indium-AI-Labs/indium-agentkit

Skill Codex

Plan and verify a staged rollout of a service change: preflight checks, canary cohorts, health gates with abort thresholds, a rehearsed rollback, and a verification report. Use when preparing or executing a deploy, canary, blue-green, or rollback plan. Every environment change, staging included, needs explicit…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Estimate the scope, effort, and risk of a technical task by inspecting the affected code, slicing the work, giving three-point ranges per slice, and combining them into an honest range with stated assumptions. Use when someone asks how big, how long, or how risky a change is. Not for producing the implementation plan…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Design event-sourced systems and CQRS read models: aggregates and events, an append-only store with optimistic concurrency, outbox publishing to idempotent consumers, projections with persisted checkpoints, snapshots, event versioning and upcasting, and erasure of personal data. Use when a domain needs the full…

Indium-AI-Labs/indium-agentkit

Skill Codex

Build or change a user-facing frontend feature (page, component, form, or client data flow) in the project's existing framework and design system, with the states it needs, WCAG 2.2 AA accessibility, and tests. Use when implementing UI from an agreed brief or API contract in Next.js, Remix or React Router, Vite, Vue…

Indium-AI-Labs/indium-agentkit

Skill Claude CodeCodex

Build simulation and hardware-in-the-loop test suites for embedded firmware: instrument-timed stimulus, fault injection (brownout, bus disconnect, sensor noise), recovery checks, and guaranteed safe teardown, running first in Renode, QEMU, or a software model. Use when firmware behavior must be verified against…

Indium-AI-Labs/indium-agentkit

Skill Codex

Run an evidence-led response to an active or suspected production incident: declare and frame it, keep a timeline, preserve evidence, test hypotheses, pick the lowest-risk reversible mitigation under explicit authority, and post status updates on cadence. Use when something is broken or degraded in production now. Not…