25 skills, $29 lifetime, one install

25 Claude Code skills. One install. Zero config.

Production-tested sequences for PR review, test gaps, migration safety, security triage, changelog generation, launch copy, and more. Drop into ~/.claude/skills/ and they work.

Buy Drills — $29 once → See all 25 drills Try the free CLAUDE.md analyzer →
Private GitHub repo delivery · Lifetime updates · 20% off with FOUNDINGRATE24
25
drills included
5
lanes covered
30s
install time
$29
lifetime, one payment
// the 25 drills, by lane

Five lanes. Five drills each.

Every drill is a focused, repeatable sequence — not a vague prompt. Each enforces scope (which files it touches), output shape (predictable formatting), and constraints (what it refuses to do).

Review 5 drills

pr-review-comprehensive
Run a 7-lens review on a pending PR. Intent, logic, tests, security, performance, API surface, rollback.
test-gaps
Find missing test coverage in a changed file. Ranked by blast radius, complexity, and recent churn.
migration-safety
Audit a DB migration for production risk. Locks, rewrites, online-DDL, rollback safety.
security-triage
OWASP top-10 scan across a diff. Every finding cited file:line with specific fix.
performance-smell
Flag N+1 queries, nested loops, unbounded accumulators on the request path.

Refactor 5 drills

refactor-extract
Find repeated logic across a module and propose concrete extraction — with rule-of-three discipline.
dead-code-sweep
Unreferenced exports, orphan files, dead routes. Confidence-ranked so you don't delete live code.
type-tighten
Upgrade any and Record<string, any> to structured types. No lateral swaps to unknown.
error-boundaries
Audit request entry points and async boundaries for swallowed errors and leaked stack traces.
env-audit
Find undocumented env vars and produce a startup validator with a staged rollout.

Documentation 5 drills

readme-sync
Regenerate the README from the current code state. Not aspirational — operational.
api-docs
Reference docs from exported functions + types. Every param, every return, every thrown error.
changelog-bot
Turn commits into a Keep-a-Changelog entry. Rewrites subject lines for end-user readability.
commit-quality
Review the last N commits. Propose rewrites; never force-push shared branches without consent.
onboarding-doc
Onboarding doc tuned for time-to-first-PR, not culture-statement aspirations.

Ops 5 drills

dependency-triage
Rank deps by bundle impact + staleness. Recommend remove / replace / upgrade / keep.
ci-audit
Find cache misses, flaky tests, oversized runners. Estimate time and cost savings.
bug-reproduce
Turn a vague bug report into a minimal repro test before proposing a fix.
postmortem-draft
Blameless postmortem from logs + timeline. Action items with named owners.
deprecation-plan
Stage a removal — announce, enable replacement, narrow usage, remove, clean up.

Launch 5 drills

launch-post-hn
Show HN post a jaded reader will actually scan. No "excited to announce."
launch-post-x
X/Twitter thread built for click-through, not likes. Earns the next tweet every time.
launch-post-release
Customer-facing release notes. What shipped, why care, what to do.
spec-to-plan
Feature spec to ordered, estimated, dependency-aware implementation plan.
estimate-effort
Three-point PERT estimate with stated assumptions and sensitivity analysis.
// 30-second install

Clone. Copy. Run.

After purchase, you get an invite to a private GitHub repo. Three commands and Claude Code picks them up on next launch.

# 1. Clone the private repo
git clone git@github.com:septimlabs-code/septim-drills.git

# 2. Install to Claude Code's skills directory
cp -r septim-drills/skills/* ~/.claude/skills/

# 3. Relaunch Claude Code and invoke any drill
claude "use the pr-review-comprehensive drill on this branch"
// why drills, not prompts

A prompt is what you type. A drill is what runs.

Prompts are raw text that ask Claude to be helpful. Drills enforce scope, output shape, constraints, and an ordered sequence. Consistent every time.

Prompt Typed fresh each time

"Review this PR for me please, check for bugs, security issues, make sure tests are there, look at the performance too..."

  • Different wording, different output each time
  • No enforcement of what gets reviewed
  • Can skip steps silently when context is long
  • No output-shape guarantee

Drill Loaded from ~/.claude/skills/

pr-review-comprehensive: 7 ordered lenses, structured output, severity-ranked findings, refuses to approve without citing 3+ verified items.

  • Same structure every invocation
  • Scope enforced (what it touches, what it refuses)
  • Output format predictable (severity + file:line + action)
  • Constraints documented and non-negotiable
// pricing

$29 once. Yours forever.

Lifetime access to this pack and every future drill added. No subscription, no usage caps, no seat licensing. Buy once.

Septim Drills — Lifetime
$29
One payment. Forever.
  • All 25 drills (5 lanes, 5 drills each)
  • Private GitHub repo invite
  • Lifetime updates — new drills added monthly
  • Use on unlimited personal + commercial projects
  • 14-day full refund if it's not what you wanted
Buy Drills — $29 once →

Use code FOUNDINGRATE24 at checkout for 20% off this week.

// faq

Questions before buying

How is this different from the Septim Prompts Pack ($9)?

Prompts Pack is 24 pre-written Claude Code prompts — text you paste. Drills are executable skills that Claude Code loads from ~/.claude/skills/ with enforced scope, output shape, and sequence. Same vendor, different surface area. Drills is for developers who've outgrown one-off prompts.

How is this different from the Septim Agents Pack ($49)?

Agents are full sub-agent identities (Atlas, Luca, Canon, etc.) with distinct voices and roles — you summon them for multi-step work. Drills are focused sequences for specific engineering tasks. You buy both if you run a solo-founder stack; buy Drills first if you're mainly using Claude Code in-session for code review and refactor work.

Will these work with every Claude Code version?

Drills use the standard ~/.claude/skills/ directory. They work with any Claude Code version that supports the skills directory pattern. If a future Claude Code update changes the skill format, we ship an updated pack at no additional charge.

Can I use these at work / on proprietary code?

Yes. The license covers unlimited personal and commercial projects for the individual purchaser. Not for redistribution or resale — if your team wants access, each developer needs their own license.

What if I'm unhappy with the pack?

Full refund within 14 days, no questions asked. Email SeptimLabs@gmail.com with your Stripe receipt.

How often do new drills get added?

Typical cadence is 1-2 new drills per month plus refinements to existing ones based on usage feedback. Every update lands in the same private repo — git pull to get the latest.