hishamkaram

release

Use when preparing, validating, publishing, promoting, or recovering claude-code-router releases, including CI readiness, live E2E evidence, changelog/release-note updates, draft GitHub releases, GHCR browser images, and Homebrew promotion.

codex-debate

Structured adversarial debate with Codex (GPT via the Codex CLI plugin) over a technical position — a design decision, architecture choice, root-cause hypothesis, migration plan, review finding, or "should we do X or Y". Codex first gives a blind independent take, then both sides exchange evidence-cited claims for bounded rounds, positions change only for new evidence, and the run ends with a written ruling, a claim ledger, and every concession on record. Use when asked to debate, challenge, stress-test, red-team, pressure-test, or get Codex to argue against a plan, decision, hypothesis, or approach; to settle a technical disagreement; or for a cross-model second opinion on a non-PR question. Not for reviewing a PR or diff (use two-model-pr-review) and never for implementing changes.

0 Updated today

Code & Development Listed

deep-plan-duo

Deep, evidence-only planning before implementation for one or more GitHub issues, pull-request review comments, a single PR or issue comment, or a plain feature request or bug report. Every claim is tagged and cited at a pinned commit and machine-checked, root causes are derived instead of symptoms, real structural fixes are scored against the tempting workaround with blast radius as a counterweight, Codex (GPT via the Codex CLI plugin) gives a blind independent diagnosis and then debates the plan for bounded rounds, and the run ends with one reviewed PR plan handed to Claude Code plan mode for approval. Use when asked for deep or thorough planning of issues or a change, "no assumptions, just facts", "a real fix, not a workaround", one PR that closes several issues, planning the fixes a PR review asked for, or a second opinion or debate with Codex on a plan before coding. Not for reviewing a diff (use two-model-pr-review), not for a bare yes/no motion (use codex-debate), and never for implementing changes.

0 Updated today

Code & Development Listed

two-model-pr-review

Two-model pull request review — runs an independent review, obtains a blind second review from Codex, reconciles both, verifies contested claims against the code, and emits one merge decision with a written audit trail. Use when asked to review a PR, diff, branch, or uncommitted local changes (working tree); for a second-opinion or cross-model code review; to adjudicate disagreeing review findings; or to decide whether a change is safe to merge. Review-only — never use to implement fixes.

0 Updated today

go-implementer

Use when implementing Go code in claude-code-router, including CLI commands, SQLite storage, secret handling, gateway code, provider adapters, sessions, and tests.

go-reviewer

Use when reviewing Go code in claude-code-router for interfaces, errors, secrets, CLI validation, SQLite behavior, concurrency, tests, and live E2E coverage.

go-test-writer

Use when writing tests for claude-code-router, including CLI tests, SQLite tests, fake provider tests, conformance tests, and live Claude Code E2E tests.

live-e2e

Use when adding, running, or reviewing live Claude Code E2E tests for claude-code-router model switching, gateway routing, tools, subagents, workflows, sessions, and compatibility behavior.

router-architect

Use when designing or reviewing claude-code-router gateway architecture, provider boundaries, SQLite schema, compatibility policy, model switching, or Claude Code feature preservation.

security-auditor

Use when handling or reviewing claude-code-router provider API keys, auth headers, local gateway exposure, secret redaction, logs, traces, SQLite storage, or fallback behavior.