ocr

A one-command workflow for turning a scanned PDF into a Word document, WeChat-ready HTML, and a record of the processing steps.

In plain words
What is it for?
Use it to process historical or other scanned PDFs, review OCR results page by page, apply proofreading, create DOCX and HTML files, and check the final outputs.
Why use it?
It coordinates preparation, OCR, human review, corrections, and output generation for scanned documents in one process.

Command

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/midnightdarling/collate/ocr
Clone the repo
git clone --depth 1 https://github.com/MidnightDarling/collate
Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 412 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00015 $0.00412
Opus 5 $0.00008 $0.00206
Sonnet 5 $0.00003 $0.00082
Haiku 4.5 $0.00002 $0.00041

Measured 2d ago against content hash 46ecdbccb583, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

ocr scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

commands/ocr.md · 27 lines

What it actually says

This is the public user path. Treat python3 scripts/run_full_pipeline.py --pdf <pdf> as an internal / debug path unless it has been proven equivalent by the same fresh-agent real-PDF gate.

The user handed over a scanned PDF and wants the full pipeline to produce publishable output:

$ARGUMENTS

Use the Task tool to dispatch the ocr-pipeline-operator subagent. Pass the PDF path as-is with a prompt that instructs:

  1. Input PDF is the path above.
  2. Follow the Canonical Workflow in agents/ocr-pipeline-operator.md:
    • Run python3 scripts/run_full_pipeline.py --pdf "<pdf>" for the mechanical stages (prep → OCR).
    • Read <workspace>/_internal/_pipeline_status.json. If status=awaiting_agent_review, build review/page_review_packets.json, classify the document type (classics / republican / modern), invoke historical-proofreader with prep/pages/*.png plus the packet file, and verify that <workspace>/review/raw.review.md is mechanically page-grounded before continuing.
    • Re-enter python3 scripts/run_full_pipeline.py --workspace "<workspace>" to chain apply-review, diff-review, to-docx, mp-format.
    • Verify final.md, previews/diff-review.html, review/diff-summary.md, output/*_final.docx, output/*_wechat.html all exist.
  3. On success, report using the Human-Facing Delivery Message template; on failure, return the structured Failure Contract.

Relay the agent's delivery message verbatim to the human — no second layer of summary. The three things they care about: deliverable paths, audit summary, residual risks.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 27 lines · 15 tokens per session scan A 46ecdbccb583

Subscribe to this mod's changes

ocr is a command published in the GitHub repository MidnightDarling/collate (6 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 15 tokens to every session and 412 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.