call-topic-mining
Skill1 starsChanged 17 days agoMITView on GitHub
Use this skill once a quarter, when the content plan needs to come from what buyers and clients actually ask rather than from a brainstorm. Produces a ranked list of recurring topics, each with the count of independent calls behind it, where in the buying cycle it comes up, and the one asset that should answer it. Also use for "what should we write about next quarter", "what do prospects keep asking", "content plan from our calls", "which objections keep killing deals".
In this repository67 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- morning-paper-outreachSkill
- client-activation-checkinSkill
- client-churn-autopsySkill
- client-win-backSkill
- late-payment-chaseSkill
- network-path-mappingSkill
- rate-increase-planSkill
- sales-call-debriefSkill
- weekly-funnel-reviewSkill
- buyer-readSkill
- punch-upSkill
- ceo-exec-assistantSkill
- og-image-from-templateSkill
- weekly-site-and-ad-reportSkill
- icp-reviewSkill
- weekly-content-ideas-from-callsSkill
- client-performance-reviewSkill
- cold-email-sequenceSkill
- email-content-reviewerSkill
- scoring-analysis-and-distributionSkill
- delivery-audit-discovery-callSkill
- directionSkill
- rundownSkill
- source-checkSkill
- work-reviewSkill
- demandbird-exact-time-linkedin-postSkill
- kondo-owed-reply-sweepSkill
- linkedin-export-contact-importSkill
- mail-messages-contact-touchesSkill
- riverside-short-social-scheduleSkill
- youtube-transcript-panel-pullSkill
- email-voice-setupSkill
- upsell-opportunity-briefSkill
- ad-creative-content-strategySkill
- klaviyo-account-auditSkill
- facilitator-notesSkill
- local-prospectingSkill
- weekly-linkedin-postsSkill
- linkedin-watchlist-briefSkill
- pipeline-by-scriptSkill
- pre-session-flash-cardSkill
- write-in-the-owners-voiceSkill
- reddit-gtm-sprintSkill
- process-captureSkill
- sales-call-market-analyzerSkill
- linkedin-shoutout-postSkill
- plain-language-copy-checkSkill
- revenue-goal-funnel-mathSkill
- case-study-buildSkill
- cash-runway-forecastSkill
- client-profitability-readSkill
- client-proposalSkill
- content-briefSkill
- quarter-goal-filterSkill
- win-to-anglesSkill
- activity-incentive-sprintSkill
- candidate-interview-coachSkill
- client-list-stack-pitchSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 8 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 42
MUST extract before clustering.
- Distinctive: line 43
MUST count independent companies, not mentions.
- Distinctive: line 44
MUST include both sales and delivery calls.
- Distinctive: line 45
MUST keep the buyer's own words from the transcript through to the asset brief.
- Distinctive: line 46
MUST strip client and company names from anything published, keeping role, size and situation.
- Distinctive: line 47
MUST compare against last quarter's list before planning.
- Distinctive: line 48
NEVER rank a topic raised by fewer than three companies, whatever it felt like on the call.
- Distinctive: line 49
NEVER answer a process problem with content.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Applies once a quarter, to a full quarter of sales and client calls at once. Produces a ranked topic list with the evidence behind each one, and the asset each should become.
The play
- Take a full quarter of calls, both kinds. Sales calls say what stops people buying. Client and delivery calls say what confuses people who already bought, which is where the authority content comes from. A quarter of one kind is half a plan. Forty calls is a workable floor; under twenty, wait another quarter rather than ranking noise.
- Extract questions and objections, not themes. The unit is one question asked or one objection raised, tagged with the company, the role and where in the cycle it happened. What qualifies and how to tag it is in references/extraction-rules.md. Theming as you read is the mistake that makes everything come out as "they want more clarity on pricing".
- Cluster the units into topics only after extraction is finished. A topic is a cluster of units asking the same underlying question in different words. The buyer's own phrasing is kept and carried through to the asset.
- Score each topic on four things: how many independent companies raised it, where in the cycle it lands, whether it has cost a deal, and whether existing content already answers it. Thresholds and the ranking matrix are in references/scoring.md. Counting companies rather than mentions matters: one talkative prospect raising a thing five times is one data point.
- Branch by where the topic lives:
- Raised before the first call, in discovery: this is top-of-funnel and belongs in public content.
- Raised at proposal or pricing: this is sales enablement first and content second. A one-pager the team can send beats a blog post nobody reads in time.
- Raised after signing, during delivery: onboarding material, and usually the highest-authority public content the firm can write, because almost nobody publishes it.
- Raised by the same client repeatedly across a project: a process problem wearing a content costume. Fix the process; writing about it will not help.
- Cut to the topics that clear the bar, then map each to exactly one asset and one internal placement. The mapping is in references/topic-to-asset.md. A topic with no owner and no date is a list item, not a plan.
- Compare against last quarter before publishing the plan. Topics that disappeared are content that worked. Topics that persist after being answered are a findability or a sales-process problem, not a content gap, and writing the piece again will not fix it.
What good looks like
- First thing I check that most people skip: the questions asked in the first five minutes of discovery calls, before anyone has framed the conversation. That is the only place the buyer's unprompted vocabulary shows up, and it is the vocabulary the content needs.
- The mediocre version: skimming a dozen recent calls, writing down the four topics that came up in the last two weeks, and calling it a content plan. It is recency, not pattern, and it produces the same four posts every agency in the category is publishing.
- I know the output is right when at least one topic is uncomfortable, meaning the honest answer costs the firm something to publish.
- Second test: the sales team recognises the list immediately and can name the deal each topic came from.
- Red flag: every topic maps to a blog post. Topics that surface at proposal stage almost never want a blog post.
Rules
- MUST extract before clustering. Never theme while reading.
- MUST count independent companies, not mentions.
- MUST include both sales and delivery calls.
- MUST keep the buyer's own words from the transcript through to the asset brief.
- MUST strip client and company names from anything published, keeping role, size and situation.
- MUST compare against last quarter's list before planning.
- NEVER rank a topic raised by fewer than three companies, whatever it felt like on the call.
- NEVER answer a process problem with content.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

