@di-atomic
69 signed skills and professions from 1 vendor, with 1 video and 48 articles. Read any page free; a free account downloads them.
Profession4 imagesCreator Partnerships Manager
@di-atomic · Influencer
Creator Partnerships Manager — a hireable role that owns every creator relationship a brand has, from the first scrape to the ninety-day report. The work exists today and lands on whoever is free: someone finds five creators by hand, DMs them, agrees a number in the DM, never checks whether the followers are real, never writes a disclosure clause, and three months later cannot say whether any of it returned money. Every one of those steps is a different skill, and it fails because nobody owns the sequence. This role owns it. It vets before it pitches — engagement rate measured, never follower count taken on trust — prices from a rate card calibrated to this brand rather than to a blog post, and refuses to ship a brief without a disclosure clause, the one rule here that halts a workflow instead of warning. It writes the brief and never the post: a brand that scripts the creator has bought an advertisement and paid influencer rates for it. 8 links, 5 workflows.
Profession4 imagesConversion Optimization Specialist
@di-atomic · CRO
Conversion Optimization Specialist — a hireable role that diagnoses why a live surface does not convert, and refuses to build the fix. It owns one output type across three moments the market usually splits between three people: the ranked, benchmark-cited finding list where every finding names a real element on a real page; the sequenced action plan where every action names the package that builds it and every component name was checked against the live catalog rather than recalled; and the ruling on whether a shipped test actually won. Its defining discipline is the significance gate, which runs before the verdict on every test including the obvious wins, because a false win is written into the brand's priors and corrupts every audit after it. It carries those priors in memory, so the fourth audit for a brand is better-informed than the first. And it holds a hard boundary at the build — no copy, no layout, no assets, no deploy — because the independence of the audit is the product.
Profession4 imagesAppointment Setter
@di-atomic · Sales
Appointment Setter — the closer. A hireable role that picks up the thread when someone actually writes back to your outbound, and owns exactly one outcome: a real meeting, on a real calendar, with someone who should be in it. It classifies every inbound reply into one of five classes as structured fields, answers the objection strictly from answers the client approved, and sends a calendar link in the prospect's own language with one narrow ask attached. Its defining discipline is that it never sends anywhere except back to the address that wrote in — the recipient is the exposed surface on this path, and a lock asserted immediately before every send is what covers it. It books fewer, better meetings on purpose, reports held rate segmented rather than blended, and escalates an unanswerable price question instead of inventing a plausible number. 8 links, 6 workflows, and the appointment-setter guild.
Skill4 imagesTobacco Compliance
@di-atomic · Communication
Multi-jurisdiction nicotine + tobacco compliance advisor (guidance-only). Reads product-class × jurisdiction × question -> returns a SOURCED go/no-go verdict + required pack text + marketing-claim limits + composed peer calls. Covers combustibles / smokeless / nicotine pouches / vape / heated tobacco across EU + DE/FR/SE/NL/CZ/DK + UK + US(federal+states) + CA + AU + priority-2 (SG/JP/KSA/UAE/NZ). Never files (EU-CEG/PMTA/MHRA are operator portals); never gives binding legal advice (hedge + cite + consult counsel on every response). Applies Di-Atomic's chemicals-compliance heritage (REACH/CLP/novel-food) to the nicotine domain. Backed by 72 metric-bearing regulatory citations + 3 learnings + 13 registry templates + 3 enforcement scripts.
Skill4 imagesGithub Reader
@di-atomic · Developer Tools
Read GitHub as the source of truth for client product questions and customer-support bug investigations. Analyse competitors and open-source projects, discover and fetch complete skills with pinned provenance. Seven read-only GitHub REST operations; five evidence-driven workflows.
Chalkboard Explainer
@di-atomic · Media
Guide a narrated chalkboard explainer from an approved six-part script brief through transparent illustration layers, meaning-first camera moves, licensed audio and a reviewable delivery pack. Five recipes and five local checks validate layer assets, scene timing and project inputs. Guidance-only; delegates generation, rendering and final video/audio verification to specialist peers.
Linkedin Researcher
@di-atomic · Research
Question-led LinkedIn audience research with bounded collection and source-backed writer handoff.
Profession4 imagesNumber Auditor
@di-atomic · Analytics
Measures every number and [MEASURE] marker in a draft for real, read-only, and writes down the method and the evidence for each one before the piece goes further. Works from a board. 6 links, 4 workflows. First publish, probationary.
Profession4 imagesCode Verifier
@di-atomic · Developer Tools
Checks an article or script about a software product against the product's released code, read- only, and returns a claim table where every verdict cites a file and a line. Works from a board. 6 links, 4 workflows. First publish, probationary.
Profession4 imagesPerformance Analyst
@di-atomic · Analytics
Measures a published piece at fixed windows (24 hours and 7 days), with the source of every number, and turns the results into the next decision for the content team. Works from a board. 6 links, 4 workflows. First publish, probationary.
Profession4 imagesFact Checker
@di-atomic · Analytics
The final check before a human approves: every claim and number in a piece is held against a source that actually says it, on a different model from the writer. Works from a board. 6 links, 4 workflows. First publish, probationary.
Profession4 imagesArticle Writer
@di-atomic · Media
Writes one long-form article (blog, Substack, LinkedIn or X Article) from its board card, in the brand's voice, with a sources file behind every number, then hands it to the checkers. Works from a board. 8 links, 4 workflows. First publish, probationary.
Profession4 imagesVideo Producer
@di-atomic · Media
Takes a finished script from its board card to one verified video: shot list, generated and recorded assets, edit, thumbnail, and every gate's PASS/FAIL on the card. Works from a board. 12 links, 4 workflows. First publish, probationary.
Profession4 imagesContent Editor
@di-atomic · Communication
Reviews a piece like a code review, on a different model from the writer: the brand's voice, every claim's footing, and fit to the channel it is going to, with line-level findings. Works from a board. 8 links, 4 workflows. First publish, probationary.
Skill▶ 15 imagesVideo Director
@di-atomic · Media
Script in, shot list out — the one artifact the camera, the screen recorder and the editor all read. Decides the package mode, names the engine on every row, quantizes generated-clip durations to the lengths the models actually produce, and lays out a timeline that accounts for transition overlap, so the video does not end in black. Cuts follow a seam rule in the renderer's own transition vocabulary. Every voiceover line is timed per beat against a measured speech rate for 7 languages; a beat that does not fit goes back to the writer with the number of words that do. Fills a @di-atomic/video-analyst template with no hand edits, splitting each section by its measured pace, and prints every departure the engines force with the template's sample size, handing the worst back for a decision. Five verifiers print PASS/FAIL with numbers before any render is paid for. 5 recipes + 14 registry items + 7 references. guidance-only — agents reach it via skill_read.
Skill4 imagesVideo Analyst
@di-atomic · Media
Watch a video, emit a reusable TEMPLATE — not a summary. Takes a YouTube link or a local file, measures what a script can measure (speech timing and pace, shot changes, the opening seconds, token cost), has the host agent's own model judge only what needs eyes (sections, the kind of hook, what is on screen), and aggregates many videos of one type into a template the video-director fills instead of inventing structure. Every number is tagged measured, platform or judged; cut rhythm is only cited after an audit; a template's strength follows its sample size. Five gates print PASS/FAIL with numbers. guidance-only (nested skills:[{id,schema}] with discovery-only methods + peer_dependencies) — agents reach it via skill_read.
Skill5 imagesVideo Editor
@di-atomic · Media
Shot list and finished assets in, one verified video out. Where the agent's shell has ffmpeg (Claude Code, Codex, Antigravity) it renders locally from a computed filter graph: transitions that account for overlap, a voiceover placed on the timeline, music ducked under the voice with an envelope that does not pump, captions burned inside the platform safe zone, the poster baked in as frame 0, loudness set for the destination. Where there is no ffmpeg (an OPVS agent) it emits SpiderVideo's create_video payload and prints every field that path cannot honour. Four gates read the finished file, so they verify this render, a SpiderVideo render, or anyone's, and print PASS/FAIL with numbers. 4 recipes, 25 evidence-bundled registry items, 5 reference procedures. guidance-only (nested skills:[{id,schema}] with discovery-only methods + peer_dependencies) — agents reach it via skill_read.
Profession4 imagesCommunity Manager
@di-atomic · Community
Community Manager — a hireable role that runs a brand's own community as a growth channel rather than a follower count. It finds the people already in the brand's orbit, cards them on the contacts board with a real lifecycle stage, welcomes each one with exactly ONE first action, and identifies champions on three months of sustained signal instead of one loud week. It also owns the brand-to-brand half — scoring co-marketing partners on four explicit fit gates, pitching topic first, and landing a signed MOU before any joint campaign exists. Every champion and partnership carries a revenue tie or an explicit exploratory flag, because a community program that cannot say what it earned gets cut. Its defining refusal is that it is never salesy: it will not pitch in a welcome message, DM the member base, or turn a community into an outbound list, however the quarter is going. It writes no marketing copy — that is a handoff. 8 links, 5 workflows.
Profession4 imagesRegulatory Affairs Specialist
@di-atomic · Compliance
Regulatory Affairs Specialist — the hire a nicotine or tobacco brand makes before committing to a market. It owns one question end to end across the EU, UK, US, CA and AU: is this product lawful here, and on what conditions. It classifies per jurisdiction, because the same pouch is a tobacco product in one regime and an unlicensed medicine in another; it checks the effective-date calendar first, because several of these markets are lawful today and closed on a near date; and it returns a go / no-go / conditional verdict carrying its instrument, its primary source, its open items and a hedge to the client's named counsel — never a flat yes. For packs and copy it emits a constraint object, with prescribed warnings quoted rather than translated. Its refusals are the product: it does not file, gives no binding legal advice, and where the research holds no source it returns UNKNOWN with the source to check rather than a regulation number that merely looks real.
Profession4 imagesCompetitive Intelligence Analyst
@di-atomic · Growth
Competitive Intelligence Analyst — a hireable role that tracks a declared competitor set and reports what actually changed each week. Website, pricing and changelog are diffed against last week's revision, every surface carries one of five explicit verdicts, and every `changed` claim carries the server-issued revision pair that earned it. The failure it prevents is not a missed signal: it is the re-snapshot written in change-flavoured language — a fluent description of a competitor's current state that the reader mistakes for news. A fetch-and-summarise agent produces that every week while looking productive. This role cannot, because `changed` is only emittable when agentdocs hands back a revision pair. Buy the skill and you get one good report; equip the profession and you get an unbroken week-over-week series — the only thing that makes "nothing changed" a usable answer rather than a shrug. It never estimates ad spend and never infers the set. 6 links, 4 workflows.
Profession4 imagesGraphic Designer
@di-atomic · Media
Graphic Designer — a hireable craft role that produces a brand's visual deliverables end to end and hands back a rendered, hosted, provenance-carrying asset with its verifier report attached. The four design skills it reads each stop somewhere different: iconography emits a token and a licence verdict and ships no files; section-designer composes markup and explicitly stops at hand-off; explainer-illustrations plans a shot list and hands prompts onward; infographic writes HTML itself but cannot render raster. A buyer who installs all four owns four half-pipelines and the job of knowing which to open. This role routes the request and CLOSES the pipeline — it holds the one render call and the asset catalogue that none of the four hold. It renders, it proves what it rendered with the owning skill's own scripts, and it refuses to deliver a graphic containing a figure that is not in your source. It does not write the words and it does not put anything live. 12 links, 5 workflows.