Develop the original ideas and IP
39 primitives in this stage — 32 skills · 7 agents. Concept-stage catalogue, kept vendor-agnostic.
Monitor the discourse on a schedule and surface what matters →
“I keep missing the conversations that matter because I only check in when I remember to.”
Standing through-line scout over the growing knowledge base →
“The connection I was looking for in January is probably already in my notes — I just never noticed it land.”
Scan the field for prior use of the term →
“The last thing I want is to build a brand around a term someone else already owns.”
Scan literature and field for prior art →
“I don't want to publish something only to be told this is just X with different labels.”
Hunt the literature and data for disconfirming evidence on demand →
“I need someone to do the search I'd skip if I were doing it myself.”
Process the capture inbox on a schedule →
“By the time I get to my inbox the notes are so old I've forgotten why I captured them.”
Traverse the store and surface unexpected connections →
“My best ideas came from notes I'd forgotten I had — but I can't keep re-reading everything to find them.”
Extract keepable quotes and reactions from a source →
“I highlighted half the page but I can never find what I was thinking when I did.”
Surface clarifying questions and pushback while reading →
“I read it, I agreed with it, and then three days later I couldn't explain why I agreed.”
Triage a feed batch into read-deep / skim / dismiss →
“My RSS feed is 400 items and I've been ignoring it for two weeks because I don't know where to start.”
Distil claims and debates from a talk or podcast transcript →
“There was something important in that episode but it's buried in 90 minutes of audio I'll never re-listen to.”
Structure a raw observation into a retrievable note →
“I had exactly the insight in that meeting and by the time I got back to my desk it was gone.”
Draft a structured reading note from markup and extracts →
“I've read it, I've annotated it, and in six months I'll have no idea what I took from it.”
Surface the candidate through-line from accumulated captures →
“I have 80 notes and I can feel there's something connecting them — I just can't see it yet.”
Sharpen the candidate thesis into an ownable, arguable claim →
“I believe this but I can't say it in a way that makes clear what's actually at stake.”
Adjudicate competing intuitions to a defensible settled position →
“I keep arguing both sides in my head and I don't know which one I actually believe.”
Draft to discover the argument →
“I don't know what I think until I write it — but writing alone doesn't always get me there.”
Curate which tentative positions to commit to and bound their scope →
“I have opinions on everything but I know I can't stake public claims on all of them — I need to decide where I actually stand.”
Generate coinage candidates →
“The idea is solid but without a name it doesn't travel — and I keep calling it something different every time.”
Stress-test a name for memorability and distinctiveness →
“I love the name but I want to know it won't turn out someone else already owns it or that it means something else in the field.”
Propose candidate structural shapes →
“I know what the idea says but I don't know if it's a 2x2 or a spectrum or something else — and that choice changes everything.”
Define the axes and check the cells carve cleanly →
“I've picked the 2x2 but when I try to name the quadrants one of them is always empty or meaningless.”
Draft the canonical short-form definition →
“I can explain it in a conversation but I've never pinned down the version someone else can cite.”
Test the definition for replicability by an outsider →
“I gave it to someone else and they applied it completely differently than I intended.”
Generate candidate anchor cases →
“I need a story that makes the whole thing click in 30 seconds — I keep using the same tired one.”
Test a candidate case against the model's structure →
“The example feels right but I'm not sure it actually demonstrates the parts I care most about.”
Decide cite-and-build vs reassert distinctiveness →
“There's something similar out there and I need to know if I should cite it, fight it, or ignore it.”
Frame a candour-inviting ask for a trusted critic →
“Every time I share a draft I get polite encouragement instead of the pushback I actually need.”
Distil candid reactions into actionable signal →
“I got five responses but I genuinely don't know which parts of the feedback to act on.”
Build an adversarial challenge agenda for the session →
“I want a session that actually tries to kill the idea, not one that ends up being a fan club.”
Capture and structure the challenges raised in the session →
“We had a great session but two days later I've forgotten what actually broke.”
Steelman the opposing position and rehearse the rebuttal →
“I know I'll face pushback at this seminar and I want to walk in having already fought the hardest version of the argument.”
Generate falsification probes and counterexample candidates →
“I keep finding evidence that supports my view — I need someone to force me to look for what breaks it.”
Triage objections into fatal / scope / weak and plan the revision →
“I have twelve objections from three different people and I don't know which ones should actually change the argument.”
Tag, status, and file a raw capture →
“My capture inbox is 300 items and every time I look at it I close the tab.”
Propose connections and a bridge note for a given note →
“I know the connection is in there somewhere — but my search never finds it because it's not the same words.”
Triage the idea backlog by readiness to develop →
“I have 40 parked ideas and I'm always developing the newest one instead of the one that's actually ready.”
Judge a note's staleness and recommend a disposition →
“I keep navigating past notes that are just wrong now and it's slowing everything down.”
Propose a restructured tagging and navigation scheme →
“My folder structure made sense two years ago and now it's fighting my thinking instead of supporting it.”