sourjya

sourjya/kiro-rails

Agent Claude Code

Performs structured code security and quality audits with restricted tool access. Supports tiered review (T1 pre-commit, T2 feature, T3 sprint).

sourjya/kiro-rails

Agent Claude Code

Adversarial verification agent. Assumes each security finding is a false positive and searches for compensating controls that disprove it. Use after the code-security-reviewer produces findings.

sourjya/kiro-rails

Agent Claude Code

Hostile UX reviewer. Assumes the implementation is flawed. Finds friction, hierarchy issues, accessibility gaps, pointer travel problems, and pattern violations. Read-only - cannot modify code.

sourjya/kiro-rails

Agent Claude Code

Performs live browser-walk UX audits using the console-idiom rubric. Restricted to read-only browser interaction - no writes, deletes, or submissions. Requires a browser MCP.

sourjya/kiro-rails

Command Claude Code

Audit the AI and agent attack surface: prompt injection, tool-permission scope, unbounded agent loops, and unsafe handling of model output. Produces an AISR report.

sourjya/kiro-rails

Command Claude Code

Audit API boundary code for contract drift: response envelope consistency, error response shapes, HTTP status-code semantics, and frontend/backend type agreement.

sourjya/kiro-rails

Command Claude Code

Audit CI/CD pipelines for missing quality gates, unpinned actions, secret exposure in logs, and deploy steps that can run without passing tests.

sourjya/kiro-rails

Command Claude Code

Structural maintainability audit: god objects, hidden coupling, dead code, comment-standard compliance, and drift from documented architecture. Produces an MRR report.

sourjya/kiro-rails

Command Claude Code

Tiered application security audit (T1 pre-commit, T2 feature complete, T3 sprint end) covering OWASP, secrets, authorization, crypto quality, and supply chain. Produces an SRR report.

sourjya/kiro-rails

Command Claude Code

Review the current branch's commits and working tree for commit and pull-request discipline, then produce well-structured commit messages and a PR description.

sourjya/kiro-rails

Command Claude Code

Audit CSS and design-token usage against the console-idiom rubric: ad hoc overrides, magic values, duplicated spacing scales, and design-system drift.

sourjya/kiro-rails

Command Claude Code

Audit dependency manifests for supply-chain risk: unmaintained or typosquatted packages, license conflicts, unpinned versions, and known vulnerabilities.

sourjya/kiro-rails

Command Claude Code

Audit frontend performance: bundle size and code splitting, render waterfalls, N+1 data fetching, unmemoized work, and lazy-loading gaps.

sourjya/kiro-rails

Command Claude Code

Scan for embedded literals that belong in config or a constants module: URLs, magic numbers, thresholds, error message strings, and environment-dependent values.

sourjya/kiro-rails

Command Claude Code

Audit infrastructure-as-code for per-environment drift, over-permissive IAM policies, unencrypted resources, and infrastructure changed outside version control.

sourjya/kiro-rails

Command Claude Code

Audit logging, metrics, tracing, and correlation IDs across pipelines and background jobs. Flags blackbox code paths that emit no telemetry and cannot be debugged in production.

sourjya/kiro-rails

Command Claude Code

Grade a Kiro spec (requirements, design, tasks) against the spec quality standards before implementation begins. Flags thin specs that will cause rework.

sourjya/kiro-rails

Command Claude Code

Audit test suite quality: coverage gaps, weak assertions, missing negative cases, over-mocking, and tests that cannot fail regardless of implementation.

sourjya/kiro-rails

Command Claude Code

Full UX audit of a frontend codebase. Produces persona cards, journey maps, heuristic sweep, anti-pattern findings, and a prioritized fix register formatted for Kiro spec output.

sourjya/kiro-rails

Command Claude Code

Live browser-walk UX review using the console-idiom rubric. Requires a browser MCP (Playwright, Chrome DevTools, or equivalent). Walks each page/route with a 9-step protocol, collects quantitative evidence, and produces a scored report with rubric IDs.

sourjya/kiro-rails

Command Claude Code

Pre-implementation UX gate. Run before coding any UI feature to validate layout decisions, interaction placement, and hierarchy. Produces a PROCEED/REVISE/BLOCK decision.

sourjya/kiro-rails

Hook Claude Code

Runs when you submit a prompt, before the agent sees it, executing branch-check.sh and session-guard.sh via bash with --merged, --format, --count, --oneline, --show-current, --porcelain and --name-only (7 commands). From sourjya/kiro-rails.

sourjya/kiro-rails

Hook Claude Code

Runs before the agent uses a tool for Bash, Edit, Write and MultiEdit tool calls, executing guard-bash.sh via bash (5 commands). From sourjya/kiro-rails.

sourjya/kiro-rails

Hook Claude Code

Runs after a tool call finishes for Edit, Write and MultiEdit tool calls, executing export-to-claude.sh via bash with --fix and --dry-run (6 commands). From sourjya/kiro-rails.