deanpeters/Product-Manager-Skills / plugins/pm-determine/skills/intel-discipline-advisor
intel-discipline-advisor
VerifiedTriage a competitive or market question into the right intelligence disciplines, cadence, and executing skill. Use when you know something needs researching but not which channel to run.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 13.0 KB |
| examples/conversation-flow-industrial.md | markdown | 4.9 KB |
| examples/conversation-flow.md | markdown | 4.9 KB |
| template.md | markdown | 2.0 KB |
examples/conversation-flow.md
View on GitHub# Intel Discipline Advisor — Conversation Flow Example
All companies, signals, and figures are **fictional**. This transcript shows the full interaction:
inline input crediting an answer (Q1 is skipped because the invocation already answered it),
adaptive questioning, honest capacity handling, numbered recommendations that show their
mapping-table logic, a combination selection, and the closing handoff to the triage card.
---
**User:** `Intel discipline advisor: our sales team lost three straight deals to Wrenchline on
"their platform is more complete" — I want to know if that objection is about to get worse.`
**Agent:** Heads-up on how this works: normally I'd ask three quick questions, but your invocation
already answered the first — this is a **suspected competitor move** (Wrenchline deepening the
platform story), with a signal already seen: three consistent win/loss objections. That's HUMINT
flagging once, from the highest-ground-truth layer there is.
**Question 2 of 3 — the follow-up:** those three losses — same segment and deal size, or spread?
(Concentrated losses suggest a targeted play; spread losses suggest a general narrative shift.)
**User:** All mid-market, all against their full-suite pitch. Also — free tools only, and honestly
I've got maybe 2 hours now and 30 min/week after that.
**Agent:** That answered **Question 3** (constraints) too — free sources, ~2 hours now,
30 min/week sustained. No more questions; here's the triage.
Your win/loss signal means HUMINT has flagged once. One discipline is an anecdote — the job is
corroboration from channels that fail *differently*. Recommendations:
> 1. **SIGINT sweep now (~1 hour, free)** — Wayback diff their pricing and product pages over the
> last two quarters; check crt.sh for new subdomains. *Mapping logic: battle cards run on
> SIGINT + OSINT + HUMINT, and SIGINT is the freshest layer — if the "complete platform" pitch
> is about to get a product behind it, staging shows up here first.* Executing skill:
> feed findings into `battle-card-builder` to rebuild the objection handling.
> 2. **OSINT review-mining pass (~1 hour, free)** — mine their reviews for whether *customers*
> echo the completeness story or contradict it ("bought the suite, use two modules"). *Mapping
> logic: the promise-delivery gap is battle-card ammunition — if the platform pitch outruns the
> product, your counter writes itself.* Executing skill: `voice-of-customer-miner`.
> 3. **Set the 30-min/week watch** — wire Wrenchline into `competitive-intel-watch`, monthly
> materiality bar, so the corroboration keeps arriving without you hunting. *Mapping logic:
> your sustained capacity is exactly the weekly SIGINT layer's cost; anything heavier would be
> cadence fantasy.*
> 4. **The off-ramp, honestly:** if the real question is "why are we losing" rather than "what
> are they building," two more win/loss interviews beat any public sweep — your churned
> prospects already know the answer. Route: `discovery-interview-prep`.
>
> Reply 1, 2, 3, 4, a combination like "1 and 3", or tell me more. (Progress: triage complete —
> recommendations on the table.)
**User:** 1 and 3. And say more about why not 2 first?
**Agent:** Good combination — 1 gives you the now-answer inside your 2 hours, 3 makes the
corroboration self-sustaining.
Why 2 isn't first *for you*: review mining is the strongest play when the decision is positioning
copy or a counter-narrative. Your stated worry is "about to get worse" — a **forward-looking**
question, and reviews look backward; they'd tell you about the platform they shipped, not the one
they're building. It's the right second sweep if option 1 finds staging evidence. That's the
mapping lesson: OSINT-voice reads the present, SIGINT reads the near future, and the question's
tense picks the discipline.
Want the one-page triage card (`template.md`) filled in from this session, so the route and its
review date land somewhere your future self will find them?
---
## Why this example works
- **Inline input was credited, twice.** The invocation answered Q1, and the user's Q2 reply
volunteered the constraints — so Q3 was never asked. Three budgeted questions, one actually
spent. Re-asking answered questions is the anti-pattern (`workshop-facilitation`,
`inline-input-flow.md`).
- **Every recommendation carries its "because."** The mapping-table logic rides along in italics —
the user leaves knowing *why* SIGINT reads the near future, which is the advisor's stated goal:
by the third session they shouldn't need it.
- **The capacity answer was believed.** 30 min/week produced a monthly-bar watch, not the
aspirational weekly deep sweep that dies by cycle three.
- **The off-ramp was offered even though the intel route was live** — and the "tell me more"
request got a real teaching answer (question tense → discipline choice), not a re-pitch.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |