session-audit SKILL.md skillA

session-audit from 5hirish/duct is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add 5hirish/duct/session-audit@git:20261110.451b27e

Pins git:20261110.451b27e in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
5hirish/duct
Path
.agents/skills/session-audit/SKILL.md
Versions recorded
5
Latest version
git:20261110.451b27e, committed Nov 10, 2026
First version recorded
Sep 19, 2026
Size
18,079 bytes
Repository stars
10
License
MIT

Independent audit of a stored Duct agent session. Pulls it from the database (the context the model read, every tool call and payload, the reply, artifact versions, memories, cost), solves the same task yourself on the same data before reading Duct's answer, grades both against a fixed rubric, diagnoses the gap to a prompt, tool, harness, model or product cause, and records proposals in a private ledger so audits compound. Use for "audit session X", "review the last brief…

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

Duct's agents run unattended on other people's data, and the only regular
reader of what they produce is the person who asked. Nobody re-derives the
answer to see whether it was right, whether the prose matched the person,
or whether the twenty tool calls were the right twenty. This skill is that
reader. It is deliberately **independent**: the auditor solves the task first,
blind, on exactly the data the agent had, and only then reads what Duct
wrote. A review that reads the answer first grades the answer's own framing.

It is also the standing comparison of Duct against other harnesses. Whoever
runs this skill (Claude Code, Codex, OpenCode, a person) is the "other
agent"; the ledger records which one, so over time it says how Duct's
insights agent compares with a frontier agent given the same inputs.

**Rubric v1.1.** Bump the version and log the change in the ledger when a
dimension is added or redefined.

## Ground rules

- **Read-only.** The pull script reads the database and never writes to it.
  Nothing in an audit changes a session, a memory or an artifact.
- **The bundle is customer data.** It lives in the private audit home and
…

Read the whole file at its exact version.

How to install

No tool to install first
claude plugin marketplace add https://markdownregistry.com/mp/a/art_geo44s5lvlzbyt4c/bcd6dadd052a3aaff52b639fa5df3cf33ab7fe0082a9d39980b1f918d43cdf51.json

Then /plugin install session-audit@markdownregistry-geo44s5lvlzbyt4c in Claude Code. It installs git:20261110.451b27e, the one graded A here, pinned to the commit it was graded at, and leaves it there.

The exact contents, by fingerprint
mdr add 5hirish/duct/session-audit@sha256:bcd6dadd052a3aaf

The line at the top of this page pins the label git:20261110.451b27e, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_geo44s5lvlzbyt4c.svg)](https://markdownregistry.com/a/art_geo44s5lvlzbyt4c)

4 badge image fetches in 30 days. What the badge shows.

Ask from your agent
claude mcp add --transport http markdownregistry https://markdownregistry.com/mcp/registry

Then ask it: what changed in 5hirish/duct/session-audit since the version I have? It reads every saved version here, no key and no account. Any agent that speaks MCP can add it.

Got a copy of session-audit from somewhere else? Ask it: where did my session-audit come from, and what was changed? It names the public original, the version yours is closest to and every changed line. The text is not kept.

Versions

The registry has recorded 5 versions of this file, between Sep 19, 2026 and Nov 10, 2026, about one every 13 days.

Watch this file (sign in, one email per new version)

versioncommittedcommitsizeaudit
git:20261110.451b27e latest2026-11-10 451b27e 18,079 BA view · diff
git:20261029.6070aea2026-10-29 6070aea 18,050 BA view · diff
git:20260929.bbe9fdf2026-09-29 bbe9fdf 18,033 BA view · diff
git:20260919.d0912ea2026-09-19 d0912ea 18,031 BA view · diff
git:20260919.9e100b42026-09-19 9e100b4 17,110 BA view
Keep it current by itself, install once

Committed this skill to a repository? Install the markdownregistry skill updates app on it, or on all of them. When the registry saves a newer version, a pull request brings your copy to it within minutes, the newest version first, then what changed and any check it newly trips. A copy you edited is never touched, and nothing is deleted.

Install on GitHub

It asks to read and write contents and pull requests, and to read metadata. No workflow file, no Actions setting, nothing runs on your runners. Uninstall it and everything the registry kept about your repository is deleted.

Keep it current by itself

Committed this skill to a repository? Add one workflow file, and when a newer version is saved, a pull request brings your copy to it, the newest version first, then what changed and any check it newly trips. A copy you edited is never touched, and nothing is deleted.

mkdir -p .github/workflows && curl -fsSL https://markdownregistry.com/ci/skill-updates.yml -o .github/workflows/skill-updates.yml

It runs daily on GitHub Actions, no key and no account. Read the workflow before you add it.

Check a whole project

From the root of your project, one line hashes each installed skill's SKILL.md, your AGENTS.md and your CLAUDE.md, and asks the registry once. Only the fingerprints and paths are sent.

sha256sum .claude/skills/*/SKILL.md AGENTS.md CLAUDE.md | curl -s --data-binary @- https://markdownregistry.com/api/v1/check-copies

In CI, add --fail-with-body: the answer is 409 when any file is behind, so the step fails and still prints which files.

sha256sum .claude/skills/*/SKILL.md AGENTS.md CLAUDE.md | curl -s --fail-with-body --data-binary @- https://markdownregistry.com/api/v1/check-copies

No sha256sum on your machine? shasum -a 256 prints the same lines.

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (18079 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

5hirish/duct · 10 stars · license MIT · pushed 2026-11-15 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_geo44s5lvlzbyt4c
GET https://markdownregistry.com/api/v1/resolve?ref=5hirish/duct/session-audit
GET https://markdownregistry.com/api/v1/blob/bcd6dadd052a3aaff52b639fa5df3cf33ab7fe0082a9d39980b1f918d43cdf51

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from 5hirish/duct

add-blog-post SKILL.md skill
5hirish/duct · .agents/skills/add-blog-post/SKILL.md · Research, write, illustrate and publish a post on the Duct blog that ranks in search and gets cited by answer engines…
git:20261110.451b27e · audit A · 10 stars
add-changelog-entry SKILL.md skill
5hirish/duct · .agents/skills/add-changelog-entry/SKILL.md · Write a new release entry on the public Duct changelog at site/changelog/ — from git history, in the voice of someone…
git:20261110.451b27e · audit A · 10 stars
add-connector SKILL.md skill
5hirish/duct · .agents/skills/add-connector/SKILL.md · Research-first evaluation of a third-party data source for Duct (OAuth, scopes, quotas, APIs) using official docs via…
git:20261110.451b27e · audit A · 10 stars
component-preview SKILL.md skill
5hirish/duct · .agents/skills/component-preview/SKILL.md · Render an app/ component in isolation at /preview to inspect layout, states, responsiveness and surfaces (dialog…
git:20261110.451b27e · audit A · 10 stars
feature SKILL.md skill
5hirish/duct · .agents/skills/feature/SKILL.md · Start a new branch safely from main. Handles uncommitted changes, stashing, unfinished merges, and branch naming…
git:20261110.451b27e · audit A · 10 stars
launch-readiness-audit SKILL.md skill
5hirish/duct · .agents/skills/launch-readiness-audit/SKILL.md · Audit a page or blog post for launch readiness across SEO, mobile responsiveness, AI crawler discoverability, and UX…
git:20261110.451b27e · audit A · 10 stars
mosaic-panel SKILL.md skill
5hirish/duct · .agents/skills/mosaic-panel/SKILL.md · Generate a Roman mosaic threshold panel for an empty state, error state, 404 or onboarding screen, and ship it as a…
git:20261110.451b27e · audit A · 10 stars
new-page SKILL.md skill
5hirish/duct · .agents/skills/new-page/SKILL.md · Create a new audience-specific landing page following the for-*.html pattern
git:20261110.451b27e · audit A · 10 stars
prioritize SKILL.md skill
5hirish/duct · .agents/skills/prioritize/SKILL.md · Product management for Duct — find what deserves attention, run it through the lean prioritization bar, file it as a…
git:20261110.451b27e · audit A · 10 stars
security-audit SKILL.md skill
5hirish/duct · .agents/skills/security-audit/SKILL.md · Run deep local security audits for backend and app changes to catch secrets exposure, DB compromise vectors, and user…
git:20261110.451b27e · audit A · 10 stars
working-backwards SKILL.md skill
5hirish/duct · .agents/skills/working-backwards/SKILL.md · Amazon-style PR/FAQ for a user-facing Duct feature. A customer press release and FAQ written on the issue before any…
git:20261110.451b27e · audit A · 10 stars
duct-sprint-refresh SKILL.md skill
5hirish/duct · .claude/scheduled-tasks/duct-sprint-refresh/SKILL.md · Manage the current sprint on GitHub Project 10 — plan it once, keep it honest on every later run, report what changed
git:20261101.83223c5 · audit A · 10 stars

Every file in 5hirish/duct

Other files named session-audit

session-audit SKILL.md skill
foyzulkarim/skills · dev-pipeline/skills/session-audit/SKILL.md · Audit Claude Code session logs (~/.claude/projects) for token waste and produce a ranked, attributed efficiency report…
git:20260904.3b8ccca · audit A · 43 stars
session-audit SKILL.md skill
ethosagent/ethos · .agents/skills/session-audit/SKILL.md · Analyze the current Claude Code session transcript for execution failures, classify errors, and suggest CLAUDE.md rules…
git:20261102.0255a79 · audit A · 23 stars

Browse by kind, by grade A, or by owner.

Kept in .agents/skills, the folder Codex reads. More agent skills for Codex.