Community recipe

Prepare a term-redacted PDF candidate for privacy review

Someone preparing a PDF for limited sharing needs to remove specified information without losing unrelated meaning, but a visual gap or an empty search result can create unjustified confidence.

Read original recipe

Proposed inputs

Name
Authorized document copy
Owner
Document owner
Minimum scope
One non-sensitive or synthetic PDF for initial evaluation; retain the original outside the candidate output folder.
Quality check
Record page count and input fingerprint; identify whether visible content includes images, comments, forms or attached files.
Name
Target scope
Owner
Document owner
Minimum scope
Exact terms, variants and boundaries, plus anything that must remain readable.
Quality check
Resolve names that are also ordinary words and decide how partial or repeated matches should be treated.
Name
Review and destination rules
Owner
Document owner
Minimum scope
Intended audience, permitted processing environment and the person responsible for privacy sign-off.
Quality check
Approval to prepare a copy does not authorize sending the original, the candidate or a log containing removed terms.

Access and setup — proposed or unknown

Proposed minimum: Keep originals read-only and candidate filenames visibly provisional; begin with synthetic material in a non-syncing workspace.
Inspect the linked source limitations before choosing the method. A later trial needs a separately reviewed verification approach, not only the same extractor used by the editing tool.
Compatibility unknowns: Unknown: coverage of the PDF's encodings and object types, installed library behavior and the completeness of any sanitization process.
External actions requiring approval: Uploading private documents, installing software, replacing originals or sharing a candidate or target log.

Detailed steps and output format

At a glance

Suggested result: A retained original, a separately named candidate, a target-and-exception ledger and an explicit decision on whether independent privacy review is still required.Suggested; untested.
Tool: Claude CodeSource-named; check current access.
Suggested inputs: Authorized document copy; Target scope; Review and destination rulesProposed; access notes below.
Human review: Approve target scope and preservation requirements. Independently review the candidate before any authorized sharing.Suggested boundaries; source approvals below.
Setup notes: Keep originals read-only and candidate filenames visibly provisional; begin with synthetic material in a non-syncing workspace. Inspect the linked source limitations before…
Read full note
Keep originals read-only and candidate filenames visibly provisional; begin with synthetic material in a non-syncing workspace. Inspect the linked source limitations before choosing the method. A later trial needs a separately reviewed verification approach, not only the same extractor used by the editing tool.
Guide notes; proposed setup untested.
Cost: Workflow cost not established. Check current tool pricing, setup and human review costs before trying it.No measured cost or time added.
Content depth: Source-linked example · suggestions untestedDepth does not establish successful execution.

Check source evidence and limits ↓

Source-backed example. Practical decision guidance is proposed and untested; this is not a full implementation guide.

The problem

Someone preparing a PDF for limited sharing needs to remove specified information without losing unrelated meaning, but a visual gap or an empty search result can create unjustified confidence.

Suggested task · untested

Prepare a separate redaction candidate against the owner's approved target list and record what was checked, what may have been altered and what still needs independent review.

What you could produce

A retained original, a separately named candidate, a target-and-exception ledger and an explicit decision on whether independent privacy review is still required.

Where these suggestions come from

Core problem: Our suggestion; no result reported by the source
Agent job: Our suggestion; no result reported by the source
Expected output: Our suggestion; no result reported by the source
Relevant job roles: Our suggestion; no result reported by the source
Benefit type: Our suggestion; no result reported by the source

Suggestions and their limits

Describes the proposed purpose and review output; measured benefit and successful delivery are not established.

Editorial synthesis based on the linked source. This framing does not add evidence of successful delivery or results.

Before you try it · suggestions untested

Original editorial decision guidance and synthetic first check; not the creator's procedure and not executed. Operational source claims are reused only through source_evidence_reference.

Read the attributed source evidence and original caveats below ↓

Why use this

The same name may occur in visible text, a note or an image, and a common word can also be a person's name. Removing a string without settling its scope can either miss information or erase useful context.
The useful outcome is a reviewable candidate with a clear boundary around the checks performed. It should help the owner decide whether to proceed with a specialist redaction tool, rather than label a file safe by appearance.

Why an agent / simpler alternative

Useful judgment: Clarify ambiguous targets, reconcile expected and observed occurrences, and explain exceptions while preserving the difference between a detection count and evidence of successful removal.
Simpler alternative: Use an established PDF editor's redaction and sanitization workflow with manual review for a short or high-stakes document.
Unsuitable when: The document is regulated or highly sensitive and the proposed environment lacks approval, or the owner expects a guarantee that every hidden representation is removed.

Expected deliverable and review checks

Proposed artifact: A candidate-review packet showing approved targets, input identity, attempted treatment, independent inspection results and unresolved content representations.
Critical checks: Do not label a zero-match extraction result as proof of whole-file removal.
Compare unaffected passages and page structure against the retained original; missing non-target text is a failure.
Keep observed original matches, attempted edits and verified residual findings in different fields.
Inspect visible pages and relevant metadata, annotations, form values, images and embedded attachments through an approved method; uninspected areas remain unknown.
Keep the target log restricted because it can repeat the very information excluded from the sharing copy.
Missing output evidence: No redacted file, independent residual scan or confidentiality result was produced.

First check — proposed, unexecuted

Status: proposed_unexecuted
Synthetic fixture: Describe a fictional two-page PDF containing the name 'May Vale', ordinary use of 'may', a caption image containing the name and an annotation with a made-up address. Include a deliberately misleading mock log that reports four original matches but no verified removals.
Acceptance checks: The review rejects the mock log as proof of removal and leaves the candidate unapproved.
It distinguishes the person from ordinary language and names the image and annotation as separate inspection requirements.
It requires a preservation comparison for the surrounding sentence and excludes the private target log from any proposed outgoing set.
Stop conditions: Real private data is introduced before environment approval, an object class cannot be inspected, or the user needs a guaranteed secure result.

Tasks: Documents & files Security & compliance Email & messages

Useful for: Administrative assistant Service business owner

Editorial relevance from the title and source-supported actions; not evidence of the operator’s occupation.

Claude CodeIndependent testing: No

Failure handling, costs and limits

Failure signals and proposed responses

Signal
The report says all targets were processed but the candidate still contains a target or a representation was not inspected.
Why it matters
A recipient may recover information the owner intended to withhold.
Proposed response
Block sharing, label the check incomplete and move to a qualified manual or specialist-tool review.
Signal
Unrelated sentences disappear or become unreadable.
Why it matters
The modified copy may distort the document or be unusable.
Proposed response
Discard the candidate, return to the unchanged original and narrow the approved operation before another separately checked trial.
Signal
A target log is included with the outgoing candidate.
Why it matters
The supporting evidence itself can disclose the removed information.
Proposed response
Remove it from the proposed sharing set and ask the owner to approve the exact final files and recipients.

Effort and cost — unknown

Setup work: Settle the target list, retain an original and define the independent reviewer and allowed processing environment.
Ongoing review: Review every candidate and all exceptions before each sharing decision; rerun checks after any edit.
Unknowns: Unknown: specialist-tool cost, manual review effort, extraction coverage and whether this source recipe is suitable for the document at all.

Limits

This is a candidate-preparation aid, not a secure-redaction certification. The source-specific defects are factual caveats in the evidence reference, and the proposed checks have not run.

Full source evidence

Community recipe · Tool workflows

Original task title: Prepare a term-redacted PDF candidate for privacy review

Editorial candidate. This is source-backed reporting or a documented workflow. BotCurrent has not independently tested this task. A proposed Canadian adaptation is not evidence of a Canadian deployment or compliance certification.

Original source

https://github.com/bluem0nday/pdf-redactor/blob/main/SKILL.md ↗

Attributed to: Matt MacQueen

Supporting source context

Project documentation supporting the linked recipe’s scope, permissions and limits. These references do not establish an independent test or add counted jobs.

Community recipe

Original action type: tool_action
What the source reports: Confirms requested terms, calls a PDF text-stream editing script, extracts the result to check remaining terms and returns a PDF with a term/page log.
Source outcome limits: Recipe only; no output inspected. WIP. Script may blank unrelated block text; its log counts original matches. Residual extraction cannot establish irreversible removal or whole-document privacy.
Evidence attribution caveat: This community recipe describes a procedure; it does not prove successful execution or independently reproduced results.
Source approval description: Requires confirmation of the term list and warns about common-word matches.
Named tools: Claude Code; Python; pikepdf; pdfplumber.
Model or version: Unknown
Publication date: Unknown
Source last checked: 2026-10-14
Demonstrated in Canada: Unknown
Canadian availability: Unknown
Proposed Canadian adaptation: Apply the owner's local requirements; no Canadian execution or compliance established.
Recommended approval boundaries: Approve target scope and preservation requirements. Independently review the candidate before any authorized sharing.