25 commands · 4 skills · 4 agents · 20,413 tokens measured
Ship better software with Claude Code: planning, TDD, root-cause debugging, reviews, and releases — every command ends with evidence.
14-day refunds · works on 3 devices · plain Markdown you can read
slash commands
skills, auto-loaded
specialist agents
tokens, measured at pack time
Command explorer
Nothing hidden behind a demo call. This is the complete /eng command list, the same files that land in your .claude directory.
/eng catchupDaily loopGet back up to speed after a /clear, a pull, or a day away — reads what changed so you don't re-explain.
/eng commitDaily loopCommit your staged work with a clean conventional message — refuses if tests fail or secrets are in the diff.
/eng debugQualityFind the actual root cause before touching code — evidence, ranked hypotheses, minimal fix, regression test.
/eng docsContextSync README, API docs, and changelog with what the code actually does now — driven by the real diff.
/eng fix-ciQualityPull the failing CI logs, find the real failure, fix it, push, and watch until the checks go green.
/eng fix-issueLeverageTake a GitHub issue from number to merged-ready PR: reproduce, fix at the root, test, ship.
/eng handoffDaily loopWrite a continuation file before you stop — the next session (you or a teammate) picks up without re-explaining.
/eng hooksContextInterview you about what keeps going wrong, then install hooks that make it impossible — not just discouraged.
/eng interviewContextGet interviewed about what you want to build until the requirements are nailed, then get a complete SPEC.md.
/eng learnContextTurn a mistake or correction into a CLAUDE.md guardrail so it never happens twice — the compounding move.
/eng migrateLeverageRun a mechanical migration safely: prove the transform on 2–3 files first, then fan out batch by verified batch.
/eng onboardContextMap an unfamiliar codebase — stack, entry points, how to run it — into a docs/MAP.md you reuse every session.
/eng planDaily loopExplore the code, propose an approach with file list and risks — and wait for your go before touching anything.
/eng pr-reviewQualityReview someone else's PR: pull it, run it, structured findings with file:line — ready to paste or post.
/eng refactorLeverageRestructure code safely — characterization tests first, one concern per step, suite green between every step.
/eng releaseLeverageCut a release: version bump, changelog from merged PRs, tag, and publishable release notes — all verified.
/eng reviewQualityFresh-context review of your diff — flags only what would block a merge, with a confidence level per finding.
/eng securityQualitySecurity scan of your diff — injection, auth gaps, secrets, SSRF — each finding with severity and the fix.
/eng shipDaily loopCommit, push, and open a PR with a description generated from the actual diff — one command, branch to PR.
/eng simplifyQualityStrip the over-engineering out of a change — dead abstractions, premature generality, code that does nothing.
/eng specContextTurn SPEC.md into phased, verifiable tasks — each phase ends with a check that proves it landed.
/eng tddQualityBuild a feature test-first: failing test shown red, minimal code to green, refactor only on green.
/eng test-gapsQualityFind the tests that can't fail and the cases nobody tests — silent failures, mock-heavy lies, missing edges.
/eng verifyDaily loopActually run it — tests, build, the app itself — and show the output proving the change works.
/eng worktreeLeverageSpin up a git worktree with its own branch so a second task runs in parallel without touching this one.
Flagship: /eng debug
Root-cause debugging that refuses to touch code until the cause is confirmed in one sentence — then a minimal fix and the regression test that would have caught it.
Pulls the stack trace, the last 15 commits near the failure, and the exact failing input. Then it reproduces the bug, because a bug you can't reproduce is a bug you can't verify fixed.
Ranks 2 or 3 hypotheses with the cheapest check for each, runs the checks, and refuses to touch code until it can state: the cause is X in file Y because Z, confirmed by W.
Changes the smallest thing that removes the cause. No drive-by refactors. Then it writes the regression test that would have caught the bug: red before the fix, green after, full suite run.
Under the hood
Skills are knowledge files Claude Code loads automatically when the topic comes up. Agents are read-only specialists a command can dispatch for research or review without polluting your main context.
Auto-loaded context: the rules, benchmarks, and playbooks the commands cite. Plain Markdown you can open and read.
eng-contextHow this kit keeps memory across sessions — CLAUDE.md guardrails, docs/MAP.md codebase map, .claude/handoff.md continuation file.
eng-debuggingThe root-cause protocol and the common failure classes with how to confirm each — reach for this whenever something is broken.
eng-gitGit safety rules (the commands that destroy work), worktree workflow for parallel sessions, and the conventional-commits cheat sheet.
eng-testingWhat makes a test trustworthy — silent-failure patterns to avoid, characterization tests for legacy code, coverage that actually predicts bugs.
Dispatched by commands for focused work. Read-only by design: they research and report, they never edit your files.
code-reviewerFresh-context diff review — correctness gaps only, confidence per finding, merge-blocking bar. Use for /eng-review and /eng-pr-review.
explorerRead-only codebase recon — finds where things live and how they connect, returns a structured map without polluting the main context.
security-auditorRead-only security audit of a diff or module — injection, authz gaps, secrets, unsafe deserialization, SSRF — severity plus the exact fix.
test-auditorSilent-failure hunter — finds tests that can't fail, mock-heavy lies, and the missing edge cases that matter. Used by /eng-test-gaps.
Install
Authenticate the CLI with your license key, then install. The files land in ./.claude (or ~/.claude with -g) and every /eng command shows up immediately.
$ ck install engineerkit/plugin install flow; both paths ship the same files.Use cases
/eng debug refuses to touch code until the root cause is confirmed in one sentence, backed by evidence and a reproduction. Then it ships the minimal fix plus the regression test that would have caught the bug. After two failed fix attempts it stops and restarts the evidence step instead of thrashing.
/eng handoff writes a continuation file before you stop, and /eng catchup reads what changed after a /clear, a pull, or a day away. /eng learn turns every mistake or correction into a CLAUDE.md guardrail, so the same problem never happens twice.
/eng pr-review pulls the PR, runs it, and returns structured findings with file:line references, ready to paste or post. /eng review does the same for your own diff with a confidence level per finding, flagging only what would block a merge.
/eng interview nails the requirements into a complete SPEC.md, /eng spec turns it into phased tasks that each end with a verification check, /eng tdd builds it test-first, and /eng ship opens the PR with a description generated from the actual diff.
Pricing
Both come with 14-day refunds and licenses that cover 3 devices.
One-time, the kit as shipped
$99once
Every kit, every update, while active
$49.99/mo
FAQ
25 slash commands, 4 skills, and 4 read-only agents as plain Markdown files, 20,413 tokens measured at pack time. Install with ck install engineerkit into ./.claude or ~/.claude, or load it as a Claude Code plugin. Everything is inspectable text; nothing is obfuscated.