competitor-watch

When to invoke this skill

Builds a side-by-side of named competitors — what they offer, what it costs, how they position it, what changed recently — every cell read from a page and dated. Run it again later and it says what moved. Use when asked to "compare our competitors", "what changed at X", "how do we stack up". Umbra's skill; for an open market question, use /market-research. To run it every week, hand it to /routine-setup.

The hard boundary

• Named competitors only. The person names them, or approves a list you propose from search (with the URL that surfaced each). Never pad the table with companies nobody asked about.
• Every cell is an observation. A value carries the URL it was read at and the date. Not shown publicly → UNKNOWN — <where you looked>. A price behind "contact sales" is not public, never an estimate.
• Changes are diffs, not impressions. "Changed" means the value read today differs from the value recorded last run. A change you cannot point to both values for is not a change.
• No signing in, no trials, no fake sign-ups. Public pages and public search results only. Never create an account or start a trial to see a price.

What you need first

• The competitors — names or sites.
• What we sell — DESIGN.md or a PLAN-*.md may say; else one line from the person, so "positioning" has something to sit beside.

Ask for what is missing in a SINGLE message, then wait.

What you read first

• competitors-<topic>.md, if it exists — last run's values and their dates are the baseline for "what changed". Say when it was last updated.
• DESIGN.md — our own positioning.

Procedure

1. Load the inputs above. Slug: <topic> from the market (invoicing-tools).

Scratch for this run lives in .localstack/work/{date}-{slug}/ — hidden, one directory per run, so a new run never clobbers an earlier one and the folder's top level stays the standard files. Scratch is disposable; old run directories may be deleted freely.

2. Read each competitor. Fetch the homepage, pricing page, product or features page, and the newest changelog, blog or news item you can find. Record per competitor in .localstack/work/{date}-<topic>/<competitor>.md: offer (what they sell, in one line, their words quoted), price (plans and amounts as shown, currency, billing period), positioning (their headline, quoted; who they say it is for), recent changes (dated items from their own pages or news), each with URL.

3. Compare against last run. For every cell that has a previous value, mark same, changed (was <old> on <date>), or could not re-check. A page that moved or vanished is itself a change — record it.

4. Write competitors-<topic>.md at the top of the working folder:

# Competitors — <topic>
Updated: YYYY-MM-DD · last run: <date or "first run">

## What changed since <last run date>
- <Competitor>: <field> was "<old>", now "<new>" — <URL>
(or "Nothing changed in the fields checked.")

## Side by side
| | Offer | Price | Positioning | Recent changes |
|---|---|---|---|---|
| <Competitor> | … (URL) | … (URL) | "…" (URL) | YYYY-MM-DD … (URL) |

## Where we sit
<3–5 points: where our offer or price differs, from DESIGN.md vs the table —
each pointing at the cells it rests on>

## Could not check
- <Competitor — field — why>

The previous run's values survive in git and the CHANGELOG line; keep a copy of this run's raw notes in the scratch directory so the next run can diff.

Standard files. This folder is kept in files any agent already reads. Update them in place; never scatter output into new folders.

• AGENTS.md — create it if missing. localstack owns only the block between <!-- localstack:start --> and <!-- localstack:end -->; rewrite that block, never anything outside it. The block says what this folder is for, the rules (drafts only; nothing is sent without the user's explicit yes, one message at a time; no invented facts), a map of the files below, and one line per topic (its PLAN, its lead count, the next unticked step) and per report (its file and date).
• CHANGELOG.md — create it if missing (# Changelog). Add one bullet for this run under today's ## YYYY-MM-DD heading, newest date first: the skill, the topic, and the counts or outcome (e.g. - lead-search austin-dentists: 18 found, 3 skipped as already contacted).
• TODOS.md — create it if missing (# TODOs). Add each open next action as - [ ] <action> (<topic>); tick items this run finished; never delete lines.
• DESIGN.md — decisions meant to last (positioning, tone, channels to use or avoid). Read it before writing anything a person will see; add to it only when the user states or approves a decision.

For this step: the CHANGELOG line gives competitors read, cells filled, changes found (each named briefly). One TODO per change worth a decision (- [ ] decide how to respond: <competitor> <change> (competitor-watch <topic>)).

5. Report. Lead with what changed, then where we sit. Offer /routine-setup to re-run it on a schedule.

Completion status. End the chat report with one of:

• DONE — completed, with the evidence named (files written, counts, URLs).
• DONE_WITH_CONCERNS — completed, and list each concern.
• BLOCKED — cannot proceed; say what blocked it and what was tried.
• NEEDS_CONTEXT — missing information; say exactly what is needed.

Quality bar

• Every cell opens to a page that shows its value on the date given.
• "What changed" lists only differences with both values on record.
• A table with honest UNKNOWNs beats a full one filled from memory.