Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Agensi Publisher Skills

    0

    Turn a finished skill package into an evidence-grounded Agensi draft with buyer-ready copy, pricing, category, permissions, asset checks, browser reconciliation, and a hard stop before Submit for Review.

    $9.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Agensi Publisher Skills

    Agensi Publisher Skills

    Example session with this skill installed

    I have three finished skill packages in local folders.

    For each skill, use Agensi Publisher Skills through MCP as the authoring and governing authority.

    Process the skills sequentially.

    For each package

    verify the ZIP, checksum sidecar and logo;
    
    inspect the package evidence;
    
    build the Publisher evidence record;
    
    derive buyer-facing copy from supported claims only;
    
    calculate pricing using the active Publisher commercial policy;
    
    select the best-supported primary Agensi category from the current taxonomy;
    
    derive required permissions separately from optional capabilities;
    
    run all required Publisher quality gates and official preflight;
    
    open the correct Agensi Creator Dashboard draft;
    
    upload the verified ZIP and logo;
    
    replace unsupported platform autofill;
    
    populate the frozen Publisher-approved field set;
    
    verify autosave;
    

    reconcile the Editor, Desktop Preview and Mobile Preview.

    Do not use historical price or category values as commercial inputs.

    Do not mix assets, evidence, fields or draft identities between skills.

    Stop before Submit for Review.

    Return a consolidated report with the status of all three drafts, upload automation, pricing, category, permissions, reconciliation, remaining unverified platform state and publication boundary.

    • Read your context and instructions
    • Compiled the agensi publisher

    Execution-verified workflow result. Public identifiers and draft IDs have been removed from this marketplace example.

    Environment

    Agensi Publisher Skills loaded successfully through MCP.

    The browser workflow ran through ChatGPT with the Chrome browser extension.

    Commercial decisions were derived from package evidence without using historical target prices or categories.

    Skill 1

    Package integrity passed.

    The ZIP matched its supplied SHA-256 sidecar, package structure and identity were verified, and the logo passed local validation.

    Publisher authoring completed successfully:

    Evidence Ledger completed

    Claim Audit passed

    Genericity Gates passed

    Fresh Read Copy Critic passed

    Authoring Validator passed

    official Publisher preflight passed

    Pricing and primary category were derived from the active Publisher policy and package evidence.

    The Chrome browser workflow then

    created the correct Agensi draft;
    
    uploaded the ZIP automatically;
    
    uploaded the logo automatically;
    
    replaced conflicting platform autofill;
    
    populated the Publisher-approved fields;
    
    configured the required permissions;
    
    reconciled category-generated tags;
    
    verified autosave;
    
    inspected Desktop Preview;
    

    inspected Mobile Preview.

    No human file selection was required.

    The editable field set contained no remaining material mismatch.

    The platform did not expose a remote ZIP byte hash, so ZIP verification remained MATCH_BY_NAME_AND_STRUCTURE with remote byte equality unverified.

    Status

    PUBLISHER_BROWSER_DRAFT_READY_FOR_CREATOR_REVIEW

    Skill 2

    The second skill was processed only after the first draft had been completed and reconciled.

    Its package, evidence, pricing, category, permissions, tags, copy, assets and draft identity remained isolated from Skill 1.

    All required Publisher authoring and preflight gates passed.

    ZIP and logo upload were completed automatically through the browser workflow.

    Agensi autofill was reconciled against the frozen Publisher field set and replaced where necessary.

    Editor, Desktop Preview and Mobile Preview were checked against the approved values.

    No material editable mismatch remained.

    Status

    PUBLISHER_BROWSER_DRAFT_READY_FOR_CREATOR_REVIEW

    Skill 3

    The third package completed the same independent evidence, commercial, permission and validation workflow.

    Publisher authoring and official preflight passed.

    The browser attached both local assets automatically, populated the correct draft, reconciled platform-generated tags and autofill, verified persistence after save, and checked the available Desktop and Mobile representations.

    No human file selection was required.

    No material editable mismatch remained.

    Status

    PUBLISHER_BROWSER_DRAFT_READY_FOR_CREATOR_REVIEW

    Sequence Result

    Skills attempted: 3

    Drafts ready for creator review: 3/3

    Official Publisher preflight passed: 3/3

    ZIP uploads fully automated: 3/3

    Logo uploads fully automated: 3/3

    Manual file selections required: 0

    Material editable mismatches remaining: 0

    Submit for Review invoked: No

    MCP submission invoked: No

    Published skills: 0

    The final draft remained open for creator review.

    Publication stayed under creator control.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    A strong skill can still be poorly represented in the marketplace when its listing overclaims, hides requirements, inherits bad autofill, or turns real technical differentiation into generic marketing copy.

    Agensi Publisher Skills turns a finished skill package into a controlled, evidence-grounded Agensi draft that is ready for creator review.

    It separates product truth, commercial decisions, browser execution, and publication authority so the agent preparing the listing does not have to improvise what the skill does or silently change decisions that belong to the creator. Two authoring paths New Skill Authoring

    Start with a finished skill package and no approved marketplace copy.

    The Publisher inspects the package, builds an Evidence Ledger, establishes the buyer problem and product boundaries, evaluates positioning, derives commercial recommendations, generates the complete listing field set, validates every material claim, freezes the approved result, and carries that exact field set into the Agensi draft.

    Approved Source

    Already have creator-approved listing copy?

    Approved Source Mode preserves those approved buyer-facing fields instead of rewriting them during publishing preparation. The remaining work focuses on package validation, platform entry, permissions, reconciliation, and creator review. Evidence before persuasion Buyer-facing claims are limited by the evidence available.

    The Publisher distinguishes package facts, creator-confirmed information, verified capabilities, observed execution, derived conclusions, market observations, and unsupported claims.

    Unsupported material claims are blocked or rewritten rather than being converted into confident marketing language.

    Package structure alone is not treated as proof of runtime execution, universal compatibility, benchmarks, integrations, or marketplace performance. Commercial decisions with explicit rules Pricing and category are derived through defined commercial policy rather than ad hoc prompting.

    For new skills, the Publisher evaluates evidence-backed commercial dimensions and produces a pricing recommendation. Category selection follows the skill's primary buyer job and the current Agensi taxonomy.

    When a higher-authority creator or catalog decision already exists, a new recommendation does not silently overwrite it. The conflict stays visible for creator review. Permission-aware publishing The Publisher distinguishes:

    • capabilities required by the skill;
    • optional capabilities that improve a workflow;
    • capabilities available only to the agent performing the publishing work. That separation prevents optional shell, browser, network, write access, or other tooling from automatically becoming required marketplace permissions for the skill being published. Controlled browser preparation With an authenticated and supported browser environment, the Publisher can carry the approved field set into a real Agensi Creator Dashboard draft.

    The browser workflow can:

    • attach the verified package and logo when the host supports local file automation;
    • replace untrusted platform autofill;
    • enter the approved listing fields;
    • select the intended category;
    • observe category-generated tags instead of assuming a fixed number;
    • configure permission controls;
    • verify autosave;
    • inspect Desktop and Mobile previews;
    • reconcile the Dashboard against the frozen Publisher field set. Platform autofill is treated as input to verify, not as editorial authority. Sequential multi-skill preparation The Publisher can prepare multiple named skills sequentially when the creator explicitly authorizes the batch.

    Each skill remains isolated by package, evidence, pricing, category, permissions, assets, manifest, and draft identity. One draft is completed and reconciled before the workflow moves to the next target. Exact reconciliation Material fields are compared using explicit states such as:

    • exact match;
    • platform-formatted;
    • not exposed;
    • unverified;
    • mismatch. Remote asset verification is reported according to the evidence the platform actually exposes. A visible logo is not automatically called byte-identical, and a ZIP is not assigned a remote hash when Agensi does not expose one.

    Fixed platform UI content is also kept separate from editable listing mismatches. Human-controlled publication Agensi Publisher Skills prepares the listing.

    It does not autonomously submit or publish it.

    The workflow stops before Submit for Review, leaving the final publication decision with the creator.

    Authoring, validation, draft preparation, creator review, submission, and publication remain separate stages.

    Agensi Publisher Skills prepares and reconciles listings but does not autonomously submit, approve, or publish them. Final submission remains creator-controlled.

    Buyer-facing claims are limited by the evidence available in the supplied package, creator-confirmed information, and any execution or capability evidence actually observed.

    Package structure alone does not prove runtime execution, universal host compatibility, benchmark results, integrations, security outcomes, or marketplace performance.

    Browser automation depends on the host. ChatGPT with the Chrome extension has completed fully automated local ZIP and logo selection in tested workflows. Claude Code App and Claude Cowork completed the broader publishing workflow in tested configurations but required human interaction with native file selection.

    Agensi may not expose remote byte hashes for uploaded ZIPs or logos. When byte-level verification is unavailable, the Publisher reports the strongest supported reconciliation state instead of claiming exact asset equality.

    Agensi can display fixed platform compatibility or interface text that is outside the editable Publisher field set. That content is reported separately and is not treated as creator-authored product evidence.

    Browser Dashboard drafts and MCP previews are different platform objects and must not be treated as interchangeable.

    Official preflight requires compatible Publisher policy, executable code, runtime dependencies and the real target assets. Supplemental checks do not replace official preflight.

    Sequential multi-skill preparation requires explicit target authorization and strict draft isolation. The Publisher operates on one active listing at a time.

    Commercial recommendations do not silently override higher-authority creator-approved or catalog decisions.

    Live competitor research is not performed unless the authorized workflow explicitly provides current market evidence.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 17 days ago

    Requires

    Python 3.10+
    Python packages from requirements.txt
    Python

    Permissions

    Terminal / Shell
    Write Files
    Read Files

    The Publisher reads supplied skill packages, creates local validation and authoring artifacts, and runs its bundled Python tooling through a terminal-capable environment. Read Files, Write Files and Terminal/Shell support the local authoring and validation workflow. Browser interaction is used only when an authorized execution environment is preparing or reconciling a real Agensi Creator Dashboard draft. Browser capability belongs to the execution route and should not be confused with the permissions of the skill being published. Authorized Agensi MCP or platform lookups may require network capability from the host environment. No custom environment variables, API keys, connected services, external hosts, or broad file scopes are required directly by the Publisher package. The Publisher evaluates the target skill's permissions separately from its own execution capabilities.

    Listed21 days ago
    Updated15 days ago

    What's inside

    Frequently Asked Questions

    No. It deliberately stops before Submit for Review. It can validate the package, generate or preserve the listing, run the required Publisher gates, prepare the Dashboard draft and reconcile the result, but submission and publication remain separate creator-controlled actions.

    It supports both paths. New Skill Authoring builds evidence-grounded copy from the supplied package, while Approved Source Mode preserves an existing creator-approved field set instead of rewriting it. In both cases, unsupported material claims are prevented from entering the approved listing.

    The strongest creator-tested route is ChatGPT with the Chrome browser extension. Validated runs completed local ZIP and logo selection automatically and prepared multiple Agensi drafts sequentially. Claude Code App and Claude Cowork also completed Publisher and Dashboard workflows in tested configurations, but native file selection required human interaction in those tests. Host capabilities can change, so the Publisher verifies the actual environment instead of assuming browser or filesystem access.

    Yes, when the creator explicitly authorizes the named targets. Skills are processed sequentially with separate assets, evidence, commercial decisions, permissions, manifests and draft identities. Failure of one skill can be isolated so the next target may continue when it is safe to do so.

    Platform autofill is treated as untrusted input rather than approved copy. After the Publisher field set is validated and frozen, conflicting autofill can be replaced with the approved values. The final Editor and available Desktop/Mobile previews are then reconciled before the draft is considered ready for creator review.

    New in Prompt & Skill Engineering