TypingMindTypingMindThe best frontend for AI
Pipeline Auditor logo

Pipeline Auditor

Community
WILLOSCAR
pipeline-auditor

Audit/regression checks for the evidence-first survey pipeline: citation health, per-section coverage, placeholder leakage, and template repetition.

Overview

PublisherWILLOSCAR
Repositoryresearch-units-pipeline-skills
Skill namepipeline-auditor
Stars
513
Forks
39
Bundled files
1
Links
Skill sourceRepository
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by WILLOSCAR on GitHub. Read the source before you install it.

Installation

Install the Pipeline Auditor AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/WILLOSCAR/research-units-pipeline-skills/tree/main/.codex/skills/pipeline-auditor
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/WILLOSCAR/research-units-pipeline-skills.git /tmp/research-units-pipeline-skills
mkdir -p .claude/skills
cp -r /tmp/research-units-pipeline-skills/.codex/skills/pipeline-auditor .claude/skills/pipeline-auditor
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Pipeline Auditor in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Pipeline Auditor on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Pipeline Auditor is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Pipeline Auditor (draft audit + regression)

Triggers & routing

  • Trigger: auditor, audit, regression test, quality report, 审计, 回归测试.
  • Use when: output/DRAFT.md exists and you want a deterministic PASS/FAIL report before LaTeX/PDF.

Purpose: a deterministic “regression test” for the writing stage.

It answers:

  • did we leak placeholders or planner talk?
  • did citation scope drift?
  • did the draft fall back to generator voice (navigation/narration templates)?
  • is citation density/health sufficient for a survey-like draft?

This skill is analysis-only. It does not edit content. For all survey-family profiles, style/citation-shape violations are blocking by default.

Inputs

  • output/DRAFT.md
  • outline/outline.yml
  • Optional (recommended):
    • outline/evidence_bindings.jsonl
    • citations/ref.bib

Outputs

  • output/AUDIT_REPORT.md
  • output/TEMPLATE_RESIDUE_SCORECARD.json

What it checks (deterministic)

A150++ citation targets (used by the auditor):

  • Per-H3: >=12 unique citations (deep: >=14).
  • Global: >=150 unique citations across the full draft (recommended target: 165; deep floor: 165).

Course-paper targets:

  • Per-H3: >=4 unique citations and >=1600 non-citation characters.

  • Global: >=24 unique citations (recommended target: 32).

  • Placeholder leakage: ellipsis (..., …), TODO markers, scaffold tags.

  • Outline alignment: section/subsection order vs outline/outline.yml.

  • Survey tables: require >=1 Markdown table for course_paper, >=2 for survey/deep (inserted by section-merger from outline/tables_appendix.md; index tables remain internal).

  • Paper voice anti-patterns:

    • narration templates (This subsection ..., In this subsection ...)
    • slide navigation (Next, we move ..., We now turn to ...)
    • unmistakable pipeline voice (this run, this workspace)
    • ambiguous terms such as this pipeline, this stage, and quality gate block only when the same sentence contains a Harness anchor such as a checkpoint, Unit ID, Harness lock, attempt ledger, or template residue; ordinary subject-matter uses remain non-blocking warnings

  • Evidence-policy disclaimer spam: repeated “abstract-only/title-only/provisional” boilerplate inside H3 bodies.

  • Meta survey-guidance phrasing: survey synthesis/comparisons should ....

  • Synthesis stem repetition: repeated Taken together, ... and similar high-signal generator stems.

  • Numeric claim context: numbers without minimal evaluation context tokens (benchmark/dataset/metric/budget/cost).

  • Citation health (if citations/ref.bib exists): undefined keys, duplicates, basic formatting red flags.

  • Citation-shape hard gate: no adjacent citation blocks ([@a] [@b]) and no duplicate keys inside one block ([@a; @a]). Mid-sentence citation ratio is >=20% for course_paper and >=30% for survey/deep.

  • Citation scope (if outline/evidence_bindings.jsonl exists): citations used per H3 should stay within the bound evidence set.

  • Whole-draft deterministic template residue: split English and CJK reader-facing prose, compare each sentence with fixed fragments from the template assets selected for this Run, and reject above the Workflow limit.

  • Template source provenance: require output/FRONT_MATTER_CONTEXT.json to record the selected front-matter assets and hashes, then require the three template-owning Skill implementations to match .harness/harness.lock.json; missing provenance, legacy locks, and repository drift block acceptance.

The JSON scorecard records the measured ratio, counts, threshold, selected asset hashes, heading-aware examples, and implementation-lock result. During normal Harness execution, Completion projects its verdict and dimensions into .harness/evaluations/ledger.jsonl, including failed Attempts, so Run Audit can expose the latest measurement instead of reducing it to PASS/FAIL. The scorecard file remains the complete evidence object; the ledger is intentionally smaller. The current 10% limit is an initial policy target. The published Survey replay completes the current 31-check contract at 0/226 residue, establishing attainability for one retained Artifact set. Clean from-scratch execution, unrelated topics, and cross-profile calibration remain open.

How to use the report (routing table)

Treat output/AUDIT_REPORT.md as a “what to fix next” router.

Common FAIL families -> responsible stage/skill:

  • Placeholders / leaked scaffolds

    • Fix: C2–C4 artifacts are not clean. Route to subsection-briefs / evidence-draft / writer-context-pack, then rewrite affected sections.

  • Missing overview tables (below the selected profile's table minimum)

    • Fix: ensure table-schema + appendix-table-writer produced outline/tables_appendix.md (>=1 course-paper table; >=2 survey/deep tables; citation-backed, no placeholders), then rerun section-merger.

  • Planner talk in transitions / narrator bridges

    • Fix: rerun transition-weaver (and ensure briefs include bridge_terms / contrast_hook), then re-merge.

  • Narration templates / slide navigation inside H3

    • Fix: rewrite the failing sections/S*.md via writer-selfloop (local, section-level) or subsection-polisher.

  • Evidence-policy disclaimer spam

    • Fix: keep evidence policy once in Intro/Related Work (front matter), delete repeats in H3 (use draft-polisher or local section rewrites).

  • Citation scope drift (out-of-scope bibkeys)

    • Fix: either (a) rewrite the subsection to stay in-scope, or (b) fix mapping/bindings (section-mapper → evidence-binder) and regenerate packs.

  • Global unique citations too low

    • Fix: citation-diversifier → citation-injector (NO NEW FACTS), then draft-polisher.

  • Intro/Related Work too thin / too few cites

    • Fix: rewrite the corresponding sections/S<sec_id>.md front-matter file via writer-selfloop (front-matter path) using dense positioning + method paragraph.

  • Whole-draft template residue above the Workflow limit

    • Fix: use each scorecard example's heading, section kind, section owner, and template owner to locate the responsible front matter, chapter lead, or H3 region; rewrite its source section, regenerate the merged draft, and rerun the auditor. Do not weaken or rename template assets to make an existing Run pass.

Prevention guidance (what upstream writers should do)

If you want the auditor to PASS without a heavy polish loop:

  • Start each H3 with a content claim + thesis (avoid narration templates).
  • Use explicit contrasts and at least one evaluation anchor paragraph.
  • Embed citations per claim (avoid trailing cite dumps).
  • Put evidence-policy limitations once in the front matter, not in every H3.

Script

Quick Start

  • uv run python .codex/skills/pipeline-auditor/scripts/run.py --help
  • uv run python .codex/skills/pipeline-auditor/scripts/run.py --workspace <workspace>

All Options

  • --workspace <dir>
  • --unit-id <U###> (optional; for logs)
  • --inputs <semicolon-separated> (rare override; prefer defaults)
  • --outputs <semicolon-separated> (rare override; defaults write output/AUDIT_REPORT.md and output/TEMPLATE_RESIDUE_SCORECARD.json)
  • --checkpoint <C#> (optional)

Examples

  • Run audit after global-reviewer and before LaTeX/PDF:
    • uv run python .codex/skills/pipeline-auditor/scripts/run.py --workspace <workspace>

Troubleshooting

Issue: audit fails due to undefined citations

Fix:

  • Regenerate citations with citation-verifier and ensure citations/ref.bib contains every cited key.

Issue: audit fails due to narration-style navigation phrases

Fix:

  • Rewrite as argument bridges (content-bearing handoffs, no navigation commentary) in the failing sections/* files, then re-merge.

Issue: audit fails due to "unique citations too low"

Fix:

  • Run citation-diversifier to produce output/CITATION_BUDGET_REPORT.md.
  • Apply it via citation-injector (edits output/DRAFT.md, writes output/CITATION_INJECTION_REPORT.md).
  • Then run draft-polisher → global-reviewer → auditor.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Pipeline Auditor AI skill do?

Audit/regression checks for the evidence-first survey pipeline: citation health, per-section coverage, placeholder leakage, and template repetition.

Why use Pipeline Auditor on TypingMind?

Because you install it once and use it with any model. Pipeline Auditor is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Pipeline Auditor in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/WILLOSCAR/research-units-pipeline-skills/tree/main/.codex/skills/pipeline-auditor. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Pipeline Auditor?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Pipeline Auditor?

As many as you like. As long as a model supports skills, you can use Pipeline Auditor with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Pipeline Auditor AI skill free?

It is published on GitHub by WILLOSCAR. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

Related AI Skills

View all
openclaw logo

Openclaw Parallels Smoke

openclaw
OrganizationPopular

Prepare, snapshot, run, rerun, debug, or interpret OpenClaw Parallels guest install, onboarding, gateway smoke, and upgrade checks across macOS, Windows, and Linux.

82.2K
391K
Instructions
View
openclaw logo

Openclaw Update

openclaw
OrganizationPopular

Route routine updates of an existing local or remote OpenClaw Gateway to its installation owner, with automatic session recovery. Defer to deployment-specific skills when available.

82.2K
391K
1 files
View
openclaw logo

Tag Duplicate Prs Issues

openclaw
OrganizationPopular

Use gitcrawl to search duplicate OpenClaw PRs/issues, group related work in prtags, and sync duplicate state to GitHub.

82.2K
391K
1 files
View
openclaw logo

Update Team Server

openclaw
OrganizationPopular

Update the operator-configured Team server through its canonical owner; trust native validation and acceptance without duplicate checks or schedulers.

82.2K
391K
Instructions
View

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇