Big Brother

    1

    Strict, evidence-based code reviews that prioritize verifiable defects and project rules over subjective style.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Big Brother

    Example session with this skill installed

    Review the change from main to feature/import. Check the repository's applicable import rules, inspect related callers and tests, and report only findings supported by code or documented requirements.

    • Read your context and instructions
    • Compiled the big brother

    Example review

    This example assumes the diff changes src/import.py and tests/test_import.py. The importer opens the output file in write mode before checking whether parsing succeeded. The test covers valid input only. The repository's CONTRIBUTING.md says imports must leave existing data intact when input is invalid.

    Findings

    [High] Validate before replacing the destination — src/import.py:42

    Type: Defect

    Evidence: import_file opens the existing destination with write mode at line 42, then parses the input at line 47. Invalid input raises after the destination has been truncated. The invalid-input case is absent from tests/test_import.py. This also conflicts with CONTRIBUTING.md's import-preservation rule.

    Impact: Importing a malformed file destroys the previous output.

    Suggested fix: Parse and validate before replacing the destination, then atomically replace it only after success; add a regression case with an existing destination.

    Coverage and limits

    Reviewed range: merge base of main and feature/import through feature/import. Inspected both changed files and the caller of import_file. No test was run; the conclusion follows from the order of file opening and parsing. Other import formats were not assessed. This is an illustrative example, not a report from a real repository review.

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

    About this skill

    Code reviews often result in bikeshedding or vague suggestions that developers ignore. This skill forces a pivot to evidence-based analysis, focusing on verifiable defects and strict project rule violations rather than personal style preferences.

    What it does

    • Range fixing accurately identifies the diff boundary between branches, commits, or PR bases to ensure no changes are missed.
    • Evidence gathering treats code and documentation as proof, requiring every finding to be backed by observable logic or repository rules.
    • Defect verification tests candidate findings against existing contracts, invariants, and data flows to confirm plausible user impact.
    • Severity grading categorizes issues from Low to Critical based on security risk, data loss potential, or functional failure.
    • Contextual auditing inspects callers and error paths outside the diff to ensure the change doesn't break upstream dependencies.

    How it works

    1. Define the boundary by stating the exact commit range or branch comparison being analyzed.
    2. Scan against rules by cross-referencing the diff with the repository's internal specifications and contribution guidelines.
    3. Verify findings by tracing execution paths or suggesting focused tests to confirm the defect is reproducible.
    4. Generate the report using a structured format that leads with high-impact actionable items.

    Frameworks & tools

    This skill is framework-agnostic and works with any codebase. It relies on internal repository files like CONTRIBUTING.md, existing test suites, and project specifications to establish the ground truth for reviews.

    Why this beats prompting it yourself

    Generic prompts often hallucinate style rules or miss deep architectural side effects. This skill enforces a specific evidentiary standard, requiring the agent to cite source lines and project rules for every claim, which eliminates subjective "nitpicking."

    Use cases

    • Auditing high-stakes PRs for security vulnerabilities or data integrity risks.
    • Enforcing project-specific coding standards defined in internal documentation.
    • Validating complex refactors by tracing data flow across the entire repository.
    • Identifying breaking changes in public API contracts or interface implementations.

    Known limitations

    Does not perform implementation tasks or general code explanations. Accuracy depends on the availability of repository documentation and the clarity of the requested diff range.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.

    Only for review tasks; skip implementation. Requires clear diff boundaries. Findings must be backed by repository evidence or project rules.

    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 6 days ago

    • Free with an Agensi account

    Permissions

    Read Files
    Terminal / Shell
    Listed6 days ago

    What's inside

    Frequently Asked Questions

    Unlike generic LLM prompts that offer subjective style suggestions, this skill mandates an 'evidence-first' approach, requiring the agent to cite specific repository rules or verifiable logic paths for every reported defect.

    This skill is framework-agnostic and language-independent; it functions effectively on any codebase as long as the agent has access to your repository's documentation and source files.

    The skill uses the provided diff range to perform 'contextual auditing,' which involves scanning files outside the immediate changes to ensure new code doesn't break upstream dependencies or existing data flows.

    To get the most accurate results, ensure your repository includes a clear 'CONTRIBUTING.md' or project specification file, as the skill uses these to establish the ground truth for your specific project rules.

    You will receive the complete skill configuration and logic optimized for leading AI agents, including specialized instructions for range fixing, defect verification, and severity grading.

    New in Code Quality & Review

    Free