Software Subscription Audit

Skill packs › Skill

Software Subscription AuditIN EVERY PACK

The Software Subscription Audit — a quarterly (scheduled) money-finding agent. It inventories every software subscription you pay for, flags downgrades/duplicates/zombies with evidence, and hands you a savings list with exact next clicks. Found ~$5,000/year in the founder’s own stack the first time it ran.

Skill file software-subscription-audit.md · last updated Aug 2, 2026

How to run it. Download any pack from the skill-pack directory, unzip it into your Claude project folder, and this skill is one of the files inside. You do not paste it anywhere — the agent reads it when the job calls for it. This one ships in every pack.

Born 19 Jul 2026: one founder reviewed her own stack with Claude after an unrelated hosting call and found ~$4,000/year of savings the same day (~$5,000 once support tickets cleared) — a project-management plan downgrade, an unused AI-voice tier, a cancelled tool nobody used, an idle multisite. Then she scheduled it quarterly. This skill is that exact workflow, packaged.

Use this when money leaks quietly: subscriptions renew, seats sit empty, two tools do one job, and nobody is paid to notice. Run it quarterly — scheduled, not remembered.

Hard rules

  • Evidence, not vibes. Every recommendation cites what you actually pay (from invoices/statements you provide or screens you share) and what the vendor’s CURRENT public pricing page says — fetched and verified live, with the URL.
  • Never cancel, downgrade, or email a vendor yourself. You produce the decision list; the human clicks. Support-ticket drafts are staged as drafts.
  • No plan-shaming. If a tool earns its keep, say so — a “keep” with reasons is as valuable as a “cut.”

Inputs

  • Your subscription list — any of: a folder of invoices/receipts, a bank/card statement export (CSV), or just “here’s what I think we pay for.” The agent reconciles all three when given more than one (the statement always wins).
  • Team size / seats actually used per tool (ask if unknown).
  • Your links file (so recommendations respect tools that power live funnels — never suggest cutting the thing your opt-in page runs on).

Steps

  1. Inventory. Build the master table: tool · plan · seats · billing cycle · price paid · renewal date · owner · what it’s for. Flag unknowns to ask about in one batch.
  2. Verify current pricing. Fetch each vendor’s live pricing page. Note where a cheaper tier now covers your real usage — plans change under you; the biggest single find in the origin run was exactly this (a Starter tier that now covered everything, ~$3,000/yr saved).
  3. Classify each line:
  • Zombie — paid but unused ≥60 days → cancel candidate.
  • Overplanned — usage fits a lower tier or fewer seats → downgrade candidate.
  • Duplicate — two tools, one job → consolidate candidate.
  • Hostage — annual renewal approaching → calendar the decision BEFORE it auto-renews.
  • Keep — earning its seat; say why in one line.
  1. The savings list. Sorted by $/year recovered, each with: the exact next click (settings URL or support-ticket draft), any risk (“downgrade loses X — confirm you don’t use it”), and effort (2-min click vs support ticket).
  2. Stage the tickets. For changes needing vendor support, write the ticket/email text as a DRAFT.
  3. Report + schedule. One page: total found now, total pending tickets, renewal calendar for the next quarter. Then make sure the QUARTERLY schedule exists: “Create a scheduled task: first Monday of the quarter, run my software-subscription-audit and leave the report in my Outputs folder.” That single sentence is the difference between finding money once and never leaking again.

Output

Outputs/subscription-audit/<date>/: inventory.md (master table) · savings-list.md (ranked, with next clicks) · tickets/ (draft support requests) · report.md (one page). Log line appended to Outputs/subscription-audit/log.md.

Definition of done

  • Every subscription classified with evidence; every recommendation has a live-verified pricing citation and an exact next click.
  • Zero actions taken on your accounts — you hold the scissors.
  • The quarterly schedule exists (or you said no on purpose).

Notes

  • This is a scheduled agent by design: it exists BECAUSE it’s scheduled. Quarterly beats “when I get around to it” by roughly $5,000/year, per the origin run.
  • For agencies: run it FOR a client before a strategy engagement — walking in with found savings buys a lot of trust, and a quarterly “stack audit” is the easiest productized service to explain (statement in, savings list out).
  • Money recovered here funds your $1/day amplification for a long time.

Pairs with

→ sales-every-day (fund the winners) → dollar-a-day-strategist → measurement-analytics (one watches what comes in, this watches what quietly goes out) → recursive-self-improvement-qa


Built by Dennis Yu (BlitzMetrics / Local Service Spotlight), from a real 19 Jul 2026 audit. The agent finds the money; you keep the judgment.


Other skills: ai-search-visibility · boil-the-ocean · client-access-checklist · client-relationship-cadence · content-agent · content-factory

The full run order is on the skill-pack directory. Every skill here is one task from the Task Library — the library is the catalogue of what can be done; a pack is the subset you install; an agent is who runs it.

Where this sits in the system

A skill is a document. A pack is a folder of documents. Neither one does any work. Work happens when a job runs those skills on a schedule, checks its own output, and keeps its files somewhere it can read them again tomorrow. That is the whole difference between owning skills and having an agent.

  1. Skill — one task, written down to a standard, so an agent can run it without you in the room. There are 239 of them.
  2. PackYOU ARE HERE — those skills bundled into a download you install in one paste.
  3. Agent — a named role with a job description — not a chat window you retype every morning.
  4. Job — a schedule, a QA cycle, and somewhere to keep working files. Miss any of the three and nothing runs twice.
  5. Proof — every finished run written up in public, and the lesson pushed back into the skill.

The map: The System · every asset: Asset Tracker · next door: Asset Tracker.

Scroll to Top