Terminal
Terminal/Skill
Tracks a named set of competitors' public announcements, pricing changes, hires and funding, and sends a dated digest of what changed for a sales team.
By Clay
Works with
npx skills add clay-run/clay-skill-creator --skill competitive-intelligence-radarThe insight: competitor intel is only actionable as a dated delta plus the move it opens — and facts must never blend into reads. The naive build re-summarizes the competitor's website and greatest-hits funding news every week: archive re-heated as "intel," the same Series B surfacing each sweep, opinion woven through fact until the sales team can't tell which sentence is sourced. This radar diffs bounded windows over public exhaust, classifies each event by what it lets YOU do (displacement window, positioning counter, roadmap tell), and keeps two registers — FACT (dated, sourced, quoted) and READ (your interpretation, labeled as such). One fabricated or stale item read aloud in a deal costs the whole digest its trust.
Nothing here ships with a value. Each one is the installer's, not the author's: ask for it, never substitute a plausible default, and where an answer does not exist say which step becomes unavailable rather than guessing. Where a default IS defensible it is named below, and using it means saying so in the output.
| Input | What the installer supplies | If it is missing |
|---|---|---|
| The competitor set | names plus canonical domains | resolve and corroborate any name given. Five is a defensible cap and must be stated: a twenty-competitor radar produces noise nobody reads |
| The angles that matter | which changes are actionable for this team: pricing, launches, logo wins, leadership churn | launches, pricing, exec changes and funding is a defensible default, stated as such |
| Window and cadence | the first-sweep lookback, then sweep-to-sweep | 30 days is defensible for the first sweep; record the sweep date in every digest |
| Audience | sellers, founders, or marketing | ask — it changes the classification, not just the wording. Sellers want displacement ammo, founders want roadmap tells |
If an answer sheet is present beside this skill, load it and ask only for what it does not cover. A partial sheet is normal; a value it is missing gets asked for on its own rather than restarting the interview. Say which values came from the sheet before using them — a sheet applied silently is a wrong field nobody catches. If there is no sheet, say nothing about sheets — the check is a file lookup, not a question, so run the interview as though the feature did not exist rather than reporting an absence. At delivery, offer to save the answers back (identifiers only — never a token or a password), private and never published — and phrase the offer so it explains itself: "want me to save your answers to a file, so the next person on your team doesn't have to answer these again?"
Run clay whoami; echo "exit_code=$?". If it fails or Clay tools are missing, run the
Clay plugin's setup skill, restart if it says to, and re-run this skill. Tell the
user which workspace you're in. Confirm the sweep arms exist in this workspace before
promising them (references/radar-arms.md — the news arm is the backbone; the hiring
and product-page arms are optional escalations with their own costs).
references/classification.md) maps events to each.Per sweep: competitors × query variants × ~1 credit (quiet queries bill the same);
optional arms priced separately in references/radar-arms.md. State the worst case,
wait for approval, then build the sweep workflow once and reuse it.
Per competitor, run the news arm with entity-anchored queries (competitor name ×
angle vocabulary), windowed to the sweep bucket. Discipline per call — the full
mechanics and traps are in references/radar-arms.md:
Each surviving event gets, in this order (rules + full taxonomy in
references/classification.md):
other goes to the digest tail,
never promoted.READ: and never cite
themselves as fact.Per competitor: act-on-now items (FACT + READ) · logged items · quiet note if quiet, then a set-level summary: sweeps run, events found by class, credits spent
(measured where the surface reports it). Unchanged/quiet competitors are one line
each — quiet is a result. Offer the standing cadence, and re-verify the competitor
set quarterly (competitors get acquired; domains change hands).
Ask: "Watch our 4 main competitors — we sell workflow automation; pricing moves and
launches matter most, monthly." Cost stated: 4 competitors × 3 angle queries ≈ 12
credits/sweep — approved. First sweep, 30-day window: 31 articles → 9 event clusters
after dedupe (one launch covered by 5 outlets = 1 item). Digest highlights:
kirivale.example — FACT: usage-based pricing tier announced 2026-07-28 (in-text date), source linked, "teams under 50 seats now start free" — READ: displacement window at their per-seat renewals; lead with switching-cost calculator. One
competitor quiet (one line). One story dropped with a note: a 2024 funding
retrospective surfacing with a fresh crawl stamp — out of window by content date.
Tail: 2 other items (conference talks). 12 credits measured. Next sweep window
starts 2026-08-12.