Delivery Spec Audit

    1

    Audits coding agent deliveries against the spec: extracts evidence of what was done, files changed, and verifications from session records, then judges coverage item by item.

    Free

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    delivery-spec-audit

    Example session with this skill installed

    Audit this delivery against the spec / check if everything requested was actually done.

    • Read your context and instructions
    • Compiled the delivery-spec-audit

    A spec coverage verdict per requirement with evidence, gap list, and out-of-scope change report.

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

    About this skill

    A delivery spec coverage auditor for coding agent tasks. When an agent task finishes (or is about to be declared complete), it deterministically extracts raw evidence of what was done, which files changed, and how many verifications ran from session records (Claude Code JSONL / OpenClaw trajectory), then judges spec coverage item by item, flagging gaps and out-of-scope changes. Works with Claude Code, OpenClaw, Codex and any agent that reads SKILL.md.

    Verification counts do not guarantee a "pass" status. Weak evidence from shell redirects requires manual review. Requires local session logs or git access.

    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

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 14 days ago

    • Free with an Agensi account

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed14 days ago

    What's inside

    Frequently Asked Questions

    Unlike generic prompts that judge code by its appearance, this skill analyzes the actual telemetry of the session. It cross-references the session logs to prove whether a developer agent actually executed specific tests or merely looked at the file, eliminating hallucinations about task completion.

    The skill is designed to parse structured trajectory data from Claude Code (JSONL) and OpenClaw. It can also analyze standard Git diffs to correlate code changes with the verification steps found in the logs.

    The primary output is a Coverage Matrix that maps your requirements to specific evidence-based verdicts (Covered, Suspected Gap, or Insufficient Evidence). You also receive a list of concrete 'Counterexamples'—scenarios the agent likely missed—and a checklist of actions to finalize the delivery.

    You will receive the core logic for the semantic audit, the `extract_spec.py` Python script for log parsing, and documentation on how to integrate the audit into your existing PR or LLM-Ops workflow.

    The skill tracks whether a test command was executed, but it cannot independently verify if the logic within a custom test suite was flawed. If an agent runs a test that doesn't actually cover the edge case, the skill will flag it as 'insufficient evidence' if the telemetry doesn't show the specific input path being hit.

    New in Code Quality & Review