Evidence Claim Auditor — Flag Unsupported Negative Findings

    1

    Flag "none found" claims in reports that lack a nearby measurement command.

    $12

    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

    Evidence Claim Auditor — Flag Unsupported Negative Findings

    Example session with this skill installed

    Check incident-report.md for negative claims that don't cite a measurement command. Use a radius of 2 lines.

    • Read your context and instructions
    • Compiled the evidence claim auditor

    Summary

    I ran scripts/evidence_claim_auditor.py on incident-report.md with --radius 2. One negative claim has no measurement command nearby and needs review. The file was not modified.

    Command

    python3 scripts/evidence_claim_auditor.py incident-report.md --radius 2
    

    Output (exit code 1)

    file#1: 1 unsupported negative claim(s)
      line 7: zero errors
    

    Review

    LineClaimCommand nearby?Result
    7"Authentication failures: zero errors in the window."NoFlagged
    9"Stale lock files: none found."Yes, line 11: find /var/run/app -name "*.lock" -mmin +30Not flagged

    The line 9 claim passed only because a command sits nearby. Please confirm that the find command really measures stale locks and that its recorded result was empty; the checker does not read command output.

    Next steps

    1. Add the command that measured authentication failures (and its result) next to line 7, or soften the claim.
    2. Confirm the line 9 evidence.
    3. Re-run the auditor until it exits with code 0.
    I ran on incident-report.md

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

    About this skill

    Scan operational reports for claims such as "none found" or "zero errors" and check that a concrete measurement command appears nearby. A lightweight deterministic review aid for incident reports, audits, and research notes.

    What it checks

    • Negative-claim phrases: none found, no results, zero errors / items / matches, not present, does not exist, not detected, no differences
    • Whether a measurement command (for example grep, rg, find, curl, git, jq, wc, stat, ls, sqlite3, python, pytest) appears within a configurable number of lines (--radius, default 4)

    How it works

    1. Point the script at one or more report or notes files.
    2. It prints each negative claim that has no command nearby, with its line number and excerpt.
    3. Exit code 1 means at least one claim needs review; 0 means none were flagged.

    Limitations

    This is a heuristic. A nearby command shows that a check was probably run, not that the command was correct or that its result supports the claim. The checker does not read or interpret command output, so a person still confirms each flagged or passed claim.

    Requirements

    Python 3.10 or newer. No third-party packages.

    Heuristic only: a command within the radius counts as evidence even if it measures something else. Does not read or interpret command output. Detects a fixed list of English negative-claim phrases.

    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 12 days ago

    Permissions

    Terminal / Shell
    Network Access
    Listed12 days ago
    Updated5 days ago

    What's inside

    Frequently Asked Questions

    Phrases such as "none found", "no results", "zero errors", "zero items", "zero matches", "not present", "does not exist", "not detected", and "no differences" (case-insensitive).

    Any plain-text file read line by line, such as Markdown reports, .txt notes, or exported logs. Several files can be passed in one run.

    Install the skill folder (SKILL.md, README.md, scripts/) where your agent loads skills. Ask your agent to review a report; it runs the script and summarizes the flagged lines. You can also run the script directly in CI or a shell; exit code 1 means something needs review.

    No. It only checks whether a measurement command appears near the claim. It does not read command output, so a person must confirm that the command measures the right thing and that its result supports the conclusion.

    Three files: SKILL.md (instructions for the agent), README.md (usage and limitations), and scripts/evidence_claim_auditor.py (the checker). No third-party source code is bundled.

    You can, but the answer may differ between runs. This script applies the same phrase list and line radius every time and returns an exit code, so it can run as a repeatable gate before a person reviews the report.

    New in Code Quality & Review