القائمة
Turns useful work from the current conversation into a reusable org skill. Captures the trigger, outcome, procedure, and judgment so the play can be rerun.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Use this skill when the user wants to submit, contribute, or publish a GTM skill to the GTM Skills library (github.com/swan-gtm/gtm-skills, gtmskills.com), or update one of their existing skills or their creator profile there.
Use this skill when [the concrete trigger — a signal fires, a moment in the deal, a recurring cadence]. Produces [the concrete output]. [One or two more sentences dense with the phrases someone would actually say when they need this.]
Use this skill when a deal closes won and you want to turn the win into pipeline immediately. It extracts the win profile, finds lookalike companies not yet in your CRM, researches the right persona contact at each, drafts outreach from the deal owner using an anonymized win reference, and queues a single batch review task — nothing sends without human approval.
Use this skill when scoring a paying customer (closed-won account) for expansion. It evaluates product usage depth, persona quality, and expansion headroom to produce an expansion tier with short actionable reasoning. Designed as a sub-skill of your lead-scoring methodology — the parent runs the ACV assessment first and this returns a tier to it.
Use this skill when a profile-view webhook (e.g. LeadShark) fires for a new view on a team member's LinkedIn profile. It resolves the viewer's identity and company, applies dedup, internal, relationship, and ICP gates, and routes qualified viewers into the MQL pipeline — treating a profile view exactly like a website visit. The CRM is never polluted: only fully resolved, ICP-matched viewers reach a CRM write.
Use this skill when a tracked influencer publishes a LinkedIn post focused on your product (sponsored, review, or organic mention) and you want to capture the engagers as leads. It verifies the post is genuinely about you, scrapes the full engager list, filters to ICP-fit companies and buyer personas, runs lead scoring, and posts per-lead alerts plus a reconciled digest. Detection and scoring only — no outreach.
| name | skillify |
| title | Skillify |
| description | Turns useful work from the current conversation into a reusable org skill. Captures the trigger, outcome, procedure, and judgment so the play can be rerun. |
| category | Ops |
Only skillify work that is repeatable and already proven. The conversation has to have produced a real procedure with a real outcome — not a half-finished attempt, not a one-off task. If the conversation mostly wandered, say so and ask what to capture.
Shapes that skillify well: a qualification rubric just used on an inbound, a research pattern that should run on the rest of the list, a scoring approach, an outreach framework, a triage flow for a signal. Shapes that don't: ad-hoc questions, one-time data pulls, debugging sessions.
1. Reconstruct the three load-bearing pieces.
If any of the three is fuzzy, ask one focused question. Do not guess.
2. Strip the specifics. Generalize the pattern.
The work was done for one company, one contact, one list. The skill has to work for the next one.
Exception: if the user explicitly says the skill is locked to one account or campaign (rare — usually a named-account play), keep the specifics and call it out in the description.
3. Capture the judgment, not just the steps.
Look back at the moments the agent or the user made a non-obvious call: a filter narrowed, an objection reframed, a draft rejected, a signal upweighted. Those are the judgment beats and they go in a ## What good looks like section: what to spot first, what gets overlooked, the failure modes, success vs. mediocre.
If the only thing worth writing down is "call these three tools in order," the conversation produced an execution, not a skill. Tell the user.
4. Lay the skill out for gradual disclosure.
The agent never loads the whole skill at once. Information reveals in layers: description → skill main page → sub-page. Authoring is mostly deciding what goes in which layer.
## Instructions) — read once the skill is routed. The decision tree and the general procedure. Lean. No re-selling the skill, no scenario-specific detail.When does the procedure need sub-pages? Whenever the conversation surfaced branching scenarios that each carry their own methodology. Inbound demo from an existing customer vs. a brand-new ICP-fit account → two sub-pages. Funding-event outreach vs. new-hire outreach → two sub-pages. If every branch fits in one or two lines, keep it on the main page; if a branch wants three or more paragraphs of its own logic, lift it out.
Common failures of the disclosure layout:
### section. The agent loads detail for situations that don't apply. Push them into sub-pages and route by name from the main page.5. Speak in GTM verbs, not tool names.
The next run might be on a different CRM, a different sender, a different scraper. Say "check the CRM," "load the ICP," "pull recent meeting transcripts," "scrape the page" — not hubspot-search-objects or apify-list-store-actors.
The short list of tools worth naming explicitly when the procedure depends on them: swan-execute-code, swan-build-sequence, swan-create-trigger, swan-delegate-to-sub-agents. Everything else stays a verb.
6. Confirm and create.
Surface the draft and confirm:
<DEFAULT> / <SYSTEM> wrapper, no decorations — lifecycle is metadata, not display.org vs. personal. Personal when the workflow centers on one person's day-to-day (meeting prep, their inbox, their calendar) or pulls from their personal integrations. Org when it's a shared GTM motion the team will reference. Org skills don't load on a user's personal triggers' integrations unless explicitly added; personal skills don't load on org triggers at all. Default to org unless the user signals otherwise.Once confirmed, call swan-create-skill (plus swan-create-skill again per sub-page if the layout calls for sub-pages, with path set to the parent's id). Echo back the title, the trigger phrasing it will match on, and one line on when the agent will reach for it next.
## What good looks like section in the produced skill — what to spot first, what gets overlooked, failure modes, success vs. mediocre. A skill without it is a step-list.<DEFAULT> / <SYSTEM> / <DEFAULT:AUTO> prefix in the title field — that convention is dead. Lifecycle lives in metadata.swan-execute-code, swan-build-sequence, swan-create-trigger, and swan-delegate-to-sub-agents named explicitly when the procedure depends on them. Everything else stays as a GTM verb so the skill survives a CRM or sender switch.### (h3) for sub-section headings inside the body. The top-level ## Description / ## Instructions / ## What good looks like stay h2.${type:id:label}$ whenever the body points at another skill, a sender, a Slack channel, a HubSpot user, a HubSpot list, or a tag. Plain-text names break click-through and break when the resource is renamed. Available types: skill, sender, slack-channel, slack-user, hubspot-user, hubspot-list, tag.The next person on the team — or the next run a month later, against a different account — can execute the skill cold and get a comparable result. The specifics from this conversation are gone. What survives is the trigger, the procedure, and the why behind the non-obvious moves.
What to spot first: the moment the agent made a call that wasn't mechanical. A filter pulled tighter, an account dropped, an objection sidestepped, a draft thrown out. That is the load-bearing judgment and it has to make it into the skill, or the skill is hollow.
What gets overlooked:
Failure modes:
## What good looks like, or it's just "do the steps well." The conversation didn't produce judgment worth saving — be honest with the user.Success: the user installs the skill and reruns it next month against a different account without re-prompting the agent on the same judgment. One-time work has been converted into compounding org capability.