More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    StarIA Dev Kit — Evidence-First Debug Gate

    1

    18 full-stack development workflows for coding agents, with evidence-first debugging, testing and code review. Includes an optional local Debug Gate validator and 45 curated frontend references.

    $29

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    StarIA Dev Kit — Evidence-First Debug Gate

    Example session with this skill installed

    Run the included disposable Debug Gate 0.2.0 cart demonstration. Show the actual decisions, failure-to-pass regression result and report location. Distinguish direct hook-contract checks from live agent-host integration.

    • Read your context and instructions
    • Compiled the staria dev kit

    Reproduced failure, scoped correction, verified regression

    Executed from the package root on 7 November 2026:

    python3 examples/debug-gate-demo/run_demo.py --output-dir /tmp/staria-demo-evidence-20261107
    

    Observed result

    {
      "version": "0.2.0",
      "demonstration": "CLI and direct hook-contract calls",
      "host_integration_verified": false,
      "premature_patch_denied": true,
      "out_of_scope_denied": true,
      "scoped_patch_allowed": true,
      "reproduction_exit_code": 1,
      "verification_exit_code": 0,
      "tests_executed": 1,
      "report": "project/debug-gate-reports/incident.html"
    }
    

    The script created a separate cart project, reproduced a TypeError caused by a quantity arriving as text, checked early and out-of-scope denial through direct hook calls, recorded source evidence, applied the scoped correction and reran the original test. JSON and HTML reports were generated locally. The demonstration refuses a nonempty output destination.

    Observed checkResult
    Premature patchDenied
    Out-of-scope patchDenied
    Evidence-approved scoped patchAllowed
    Original testFailed before, passed after
    Host integration certificationNot established by this direct demo

    Separately, the full validator suite observed 96 tests in 34.981 seconds, OK, on Linux with Python 3.13.5. The optional pytest adapter was exercised in a separate test environment. The screenshots in the description and gallery are unedited captures from these actual executions.

    Next: follow the included quickstart and perform the actual host acceptance test with enabled, trusted hooks in a disposable project. Loading through Agensi does not activate local hooks. This evidence validates the tested workflow decisions, not universal host compatibility or the semantic truth of arbitrary diagnoses.

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

    About this skill

    StarIA Dev Kit brings 18 focused development workflows together with an evidence-first debugging process for coding agents working on real projects.

    Version 0.2.0 replacement update: submitted for platform review with this new ZIP. Buyers receive the replacement after approval; the currently approved package remains live during review.

    What the kit helps you do

    • Plan and implement frontend, backend, APIs, authentication and database changes.
    • Investigate bugs through observed failures, source inspection and evidence-linked hypotheses.
    • Review code, write tests and check accessibility, performance and application security.
    • Select references from a curated catalog of 45 frontend tools. These are references, not bundled third-party products, licenses or subscriptions.

    The package contains 18 task modules plus a routing SKILL.md entry point. The router is not an extra task module.

    Debug Gate 0.2.0: make the result assessable

    The optional local gate ties patch approval to a reproduced failure, an executed diagnostic, an inspected source line and a selected file scope. Verification distinguishes a verified result, cancellation and closure without proof.

    The updated checks reject weakened test definitions, zero executed cases, skipped cases, changed effective test configuration, interrupted runner results and stale-bytecode success. Changing only Git HEAD is not a scoped correction.

    Structured automated verification currently supports Python -m unittest and optional -m pytest. Other runners can provide diagnostic information but do not receive the new automated proof guarantee. Prepare regression tests before capture; changing tests requires a fresh failed capture before applying the correction.

    See real execution evidence

    These are unedited terminal screenshots captured on 7 November 2026, not generated mockups.

    A real failure-to-pass demonstration

    Actual disposable Debug Gate demonstration The disposable cart demo reproduced a TypeError, checked premature and out-of-scope denial, allowed a scoped correction, reran the original test successfully and generated a local report. These are CLI and direct hook-contract calls. The output explicitly reports host_integration_verified: false.

    96 automated tests passed

    96 automated tests passed on Linux Observed on Linux with Python 3.13.5: 96 tests in 34.981 seconds, OK. The optional pytest adapter was exercised using a separate test environment. This establishes the tested validator behavior, not universal agent-host compatibility.

    Checks against misleading success

    Automated negative-case regression checks The selected regressions confirm that weakened tests, an empty test selection and stale bytecode cannot certify a correction.

    Included with the 0.2.0 replacement

    • Local doctor/challenge diagnostic separating configuration from a correlated denied hook call.
    • Local JSON/HTML incident reports, without raw test output, command arguments, absolute project paths or explanation text. File names may still be confidential.
    • Runnable disposable demonstration, quickstart, regression tests and host acceptance instructions.

    Requirements and limits

    Workflow use requires a compatible SKILL.md agent with access to the resources and project. The optional local gate/demo requires Python 3.11+ and Git; ripgrep can be used for diagnostics. Pytest is optional and must be installed in the selected interpreter. Mechanical blocking additionally requires supported, enabled, trusted local hooks.

    Loading the skill through Agensi does not install or activate local hooks. The gate is an operational guardrail, not a security sandbox or proof that a root-cause explanation is correct. Validate host integration on a disposable project before relying on it. Native Windows acceptance was not performed in the observed screenshot session.

    The kit has no paid model API of its own. Your agent plan, project dependencies and optional external services have their own requirements and costs. Instructions include English; supporting documentation includes Portuguese. The quickstart is included in the package.

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

    The 0.2.0 replacement is subject to platform approval. Local blocking requires supported, enabled, trusted hooks; Agensi loading does not activate them. Structured automated verification currently supports Python unittest and optional pytest; other runners remain unconfirmed. Tests/configuration must stay unchanged after capture. The gate checks evidence and workflow, not root-cause truth, and is not a security sandbox. The 96-test suite passed on Linux; direct hook-contract checks do not certify live host integration or native Windows acceptance. The kit includes workflow instructions and references, not third-party licenses, hosting, API credits or guaranteed outcomes.

    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

    Money-back guarantee

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

    Trust & safety

    Security scan

    Security scan passed 10 days ago

    Software
    Python unittest
    orpytest
    Local agent hooksProject build toolsProject test tools
    Optional
    Python 3.11+Gitripgrep

    Python 3.11+ and Git run the local Debug Gate and demo; ripgrep is for diagnostics. unittest or pytest handles structured verification. Hooks must be enabled and trusted.

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access
    Environment Variables
    Browser

    Access is limited to the user-selected project and authorized tasks. Terminal and file access support inspection, tests and requested code changes. Browser/network access may be used for current documentation and authorized UI checks. Environment variables are used for project configuration and optional local gate settings; no credentials should be collected or disclosed. The optional local gate stores workflow state in ~/.staria-debug-gate/ or the configured STARIA_DEBUG_GATE_HOME. External accounts, deployments and production mutations require separate authorization.

    Listed10 days ago

    What's inside

    Frequently Asked Questions

    18 task-focused workflow modules, one routing entry point, 45 curated frontend references and the optional Debug Gate validator. The nineteenth SKILL.md is the router. The 0.2.0 replacement adds the disposable cart demonstration, quickstart, structured regression evidence, doctor/challenge and local JSON/HTML reports; buyers receive the replacement after Agensi approval.

    No. Hosted loading provides the workflow instructions. Mechanical blocking additionally requires a compatible local runtime with the hooks installed, enabled and trusted. Validate that integration on a disposable project before relying on it.

    The kit itself requires no paid model API. You need a compatible coding agent and the normal tools for your project. Agent plans, optional external services, hosting and any APIs you choose for the application have their own terms and costs.

    Agents that support SKILL.md instructions and can access the relevant module resources and project files can use the workflows. Shell commands and optional local hooks require the appropriate host capabilities and permissions. Compatibility of mechanical hook enforcement must be checked in the actual runtime.

    The 0.2.0 validator suite passed 96 automated tests on 7 November 2026 on Linux with Python 3.13.5, including an optional pytest adapter exercised in a separate test environment. Real terminal captures show the full suite, the disposable failure-to-pass demo and selected negative-case regressions. These are CLI and direct hook-contract checks, not certification of live agent interception, native Windows behavior or semantic root-cause correctness.

    No. It is a curated catalog of 45 external tool references. Check each vendor’s current license, availability and prices before using a referenced product. The package does not include paid assets, subscriptions or licenses to third-party tools.

    Structured evidence is implemented for Python -m unittest and Python -m pytest, with pytest installed in the selected interpreter. Other accepted runners can provide diagnostic information but do not receive the new automatic proof guarantee. Prepare the regression test before capture; changes to tests or effective configuration require a fresh failed capture before correcting the source.

    New in Testing & Debugging