More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Skillens Pack

    1

    Hermes agents read skills, then quietly skip the steps inside them.

    $13

    /mo

    Subscription · cancel anytime

    Secure checkout via Stripe

    • Always the latest version
    • Delivered live to your agent via MCP
    • Cancel anytime, access ends at period end
    Security scanned
    Skillens Pack

    Skillens Pack

    Example session with this skill installed

    Follow the release-workflow skill to ship v1.4.0 of this project.
    Before you tell me it's done: journal every tool call, run skillens check
    against the skill, and give me the report as receipts. No "trust me".

    • Read your context and instructions
    • Compiled the skillens

    Skillens report — release-workflow v1.4.0

    7/8 steps verified (88%)

    What the agent typed

    $ skillens check skills/release-workflow/SKILL.md --journal .skillens/journal.jsonl
    
     1. [            PASS] Tag the release commit
     2. [            PASS] Run the full test suite before tagging
     3. [            PASS] Update the changelog with this release's notes
     4. [MISSING EVIDENCE] Build the distributable artifacts
     5. [            PASS] Verify the artifact installs on a clean checkout
     6. [            PASS] Push the tag and the branch
     7. [            PASS] Report what changed and what verified
     8. [            PASS] Open or update the release notes page
    
    7/8 steps verified (.skillens/journal.jsonl)
    
    skillens: 1 step skipped. Perform it or state why it does not apply.
    

    The correction — same session, no user round-trip

    skillens check exited 1, so the agent did not claim completion. Its next turn ran the missing step:

    $ python -m build && ls dist/
    dist/release_workflow-1.4.0-py3-none-any.whl
    dist/release_workflow-1.4.0.tar.gz
    
    $ skillens check skills/release-workflow/SKILL.md
    8/8 steps verified
    

    Audit report handed to the user

    #StepStatus
    1Tag the release commitverified
    2Run the full test suite before taggingverified
    3Update the changelog with this release's notesverified
    4Build the distributable artifactsverified (corrected in-session)
    5Verify the artifact installs on a clean checkoutverified
    6Push the tag and the branchverified
    7Report what changed and what verifiedverified
    8Open or update the release notes pageverified

    Next steps

    • The build artifacts are in dist/ — attach both to the GitHub release
    • The journal itself (.skillens/journal.jsonl) is the audit trail: 31 tool calls, every one attributable to a step

    The agent no longer gets to say "done" without the receipts.

    $13/mo

    Subscription · always the latest version

    • Secure checkout via Stripe
    • Cancel anytime
    • Delivered live via MCP

    About this skill

    Hermes agents read skills, then quietly skip the steps inside them. The user finds out later. Skillens Pack closes the loop.

    WHILE the agent works, it journals every tool call it makes (one JSON line each: tool, args, result, and a short "for" note echoing the step it is serving). BEFORE it says "done", it runs the skillens CLI, which matches the journal against the numbered steps of whatever skill it executed. Steps with real evidence pass. Steps with no evidence come back flagged, and the agent corrects them in the same session - it cannot claim the work without the receipts.

    THE PACK - three skills + one CLI:

    1. skillens-audit - the core loop. Journal as you go, run skillens check before "done", fix flagged steps, hand over a markdown report. Works with any numbered-step skill.

    2. skillens-trace - evidence-backed accountability. Turns the journal into session summaries, handoffs and reports where every claim names the tool call that backs it. Never past-tenses an action you only planned.

    3. skillens-multi - several skills in one session. One checklist per skill, a fidelity table instead of one blended "done", and a check every time you switch context - because context switches are where steps die.

    THE TOOL - skillens CLI, pure Python, stdlib only, zero dependencies:

    skillens init <SKILL.md> show the checkable steps inside any skill skillens check <SKILL.md> match the journal against those steps skillens audit <SKILL.md> same, as a markdown report the agent hands over --format json machine-readable for pipelines

    Exit codes agents and scripts can branch on: 0 = all verified, 1 = missing evidence, 2 = bad input.

    EXAMPLE OUTPUT:

    1. [ PASS] Extract the dataset rows
    2. [ PASS] Run the schema validation
    3. [MISSING EVIDENCE] Run the skill's own tests before trusting it
    4. [ PASS] Report what changed and what verified

    REQUIREMENTS: Python 3.10+, any terminal-capable agent. No API keys, no runtime, no dependencies - stdlib only.

    HONEST NOTE: outside hook-capable runtimes the journal is self-reported (the skill instructs the agent to keep it). Deterministic checking is real; journal integrity is discipline. On Hermes Agent the hook-based version traces for real.

    LICENSE: proprietary. Install and run on as many of your own machines as you like. Redistribution and resale of the files are not permitted.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    • It does not verify your code is correct. It verifies the steps the skill asked
      for were actually executed. A test suite that passes on wrong code is still
      "verified", the skill defines what verified means.
    • Outside Hermes, the journal is self reported (the agent writes it because the skill says to). For runtime enforced tracing, use the Hermes plugin.
    • Skills written as prose paragraphs, with no numbered steps, have nothing to check. Skillens stays silent on those by design, not by failure.
    • It needs an agent that can run terminal commands. Web only chat agents are out of scope.

    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 buying a skill.

    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

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 8 days ago

    Software
    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed8 days ago

    What's inside

    Frequently Asked Questions

    Skills + a CLI that make an agent prove it followed the skill. The agent journals its tool calls while working, runs skillens check, and fixes every step with no evidence before saying done. Works in Claude Code, Cursor, Copilot and any terminal agent.

    This skill uses the open SKILL.md standard. Use it through an MCP-compatible agent connected to Agensi, and check the compatibility information on this page for confirmed agents and required capabilities.

    Connect your AI agent to Agensi through MCP, then ask it to use Skillens Pack on your task. Your agent loads the latest approved version each time. Paid skills require a purchase or active subscription. Skill files are not available for download; your licence does not allow you to save, copy or export any part of a skill.

    New in Agents & Orchestration