Protein Research Replay & Ranking Audit

    1

    Seven skills plus source for campaign replay, existing-score audits and structured research-decision evaluation.

    $19.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Protein Research Replay & Ranking Audit

    Example session with this skill installed

    Illustrative scenario: Review a synthetic historical campaign and summarize its existing-score audit.

    • Read your context and instructions
    • Compiled the protein research replay

    ILLUSTRATIVE FORMAT ONLY — no campaign replay, score calculation, model call or biological experiment was performed for this example.

    Review scope
    Use a synthetic historical campaign and supplied precomputed scores to demonstrate an audit record. Do not create new sequences or propose experimental work.

    Proposed audit record
    Recorded events: list the supplied decisions and timestamps with their provenance.
    Evidence gaps: flag missing inputs, unavailable pinned source files and incomplete historical records.
    Ranking comparison: distinguish the originally recorded ordering from a recomputed ordering of the existing scores, documenting assumptions only after actual analysis.
    Missing-data warnings: leave unknown values unresolved rather than filling them with plausible numbers.
    Limits: predicted or historical scores do not establish experimental binding or validate the safety of an arbitrary target.

    Review boundary
    This illustration claims no generated artifact, successful replay or measured improvement. Optional LLM evaluation may share supplied inputs with a buyer-selected provider and incur charges; obtain authorization and human risk review first. Declarative handoff packages do not execute external compute, synthesis or wet-lab procedures.

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

    About this skill

    A research workflow toolkit for structured protocols, campaign-history replay, audits of supplied precomputed scores, optional LLM decision evaluation, and declarative external handoff/import. Includes seven original skills, a root router, complete application source, schemas, tests/examples and attribution notices.

    This is not an autonomous protein design or generation tool. It contains executable ranking analysis of existing scores, but does not generate new sequences, run structure-prediction models, order synthesis, automate wet-lab work or execute external compute. External execution/data sharing requires separate approval and human risk review; arbitrary target safety is not automatically established.

    Free upstream source: https://github.com/Hylouis233/open-protein-binder-design/tree/1da3110175f0f6d885de8157c441fe18f05c4118. Original code/documentation is MIT. Two existing Anthropic API metadata fixtures retain CC-BY-4.0 attribution under THIRD_PARTY_NOTICES.md; they are not relicensed MIT. A purchase does not restrict original open-source rights or grant exclusive ownership.

    Requires Python 3.10+, pydantic and PyYAML. Optional OpenScience provides full workflow contracts. Optional networked LLM evaluation requires your own provider configuration and may incur charges. No API key, API credits, subscriptions, protein models, GPU service, hosted service or third-party dataset rights are included.

    Agensi runtime compatibility is unverified. The complete v2 security patch passed 343 local pytest cases and 39 security subtests, with repository, plugin and syntax checks documented in the package. Two network integration tests were not run. Real provider APIs, Windows and Agensi execution remain unverified; passing local tests is not an absolute security guarantee. Existing examples are upstream artifacts, not newly measured results. Availability of real-source data remains subject to the documented limitations.

    Price: US$19.99 one-time purchase for the complete package, not a subscription.

    No new protein generation or autonomous biological execution.
    Audits supplied precomputed scores; predictions do not establish experimental binding.
    Optional LLM evaluation is networked and may be billable.
    No automated guarantee that arbitrary supplied targets are safe.
    Real-source files at pinned revisions may be unavailable.

    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 1 month ago

    Requires

    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access
    Environment Variables

    Read/write user-authorized local research files. Optional bounded source-metadata/file retrieval after authorization. Optional transmission of evaluation inputs to buyer-selected LLM provider after authorization. Export/import declarative data packages; external execution requires separate approval. Scope all access to the buyer-authorized task. Configure any optional tools separately.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    The complete v2 package retains all 173 prior files and all seven original skills, with five additional files (178 files total; eight SKILL.md entries including the root router). Active attack examples were replaced with harmless markers and explicit threat descriptions, not encoded or hidden. The patch adds export path/link checks, schema type constraints, separation of external data from system instructions, and actual network response-size limits. Local validation passed 343 pytest cases and 39 subtests; two network integration tests were not run. No real API, Windows or Agensi runtime validation is claimed. The scope remains existing-history replay and supplied-score auditing, without protein generation or autonomous experiments.

    New in Research & Analysis