How to Combine Claude Cowork and the Brand-Guidelines Skill to Build a Monthly Client Recap

If Claude Cowork is new to you

Cowork runs in Claude Desktop and can work with the local files you choose to share. In this workflow, Cowork is the evidence-gathering layer. It helps inspect the client folder, summarize what was actually delivered, and flag missing support before the recap is written. For broader setup context, see the Claude Cowork tool hub and the Anthropic Workflows series.

If Claude Skills is new to you

Claude Skills are reusable folders that Claude can load when a task matches the Skill description. A custom Skill usually includes a Skill.md file with metadata, instructions, examples, and optional supporting files. Anthropic's current help says Skills are available on Free, Pro, Max, Team, and Enterprise plans when code execution is enabled.

For this workflow, you will create one private Skill, upload it in Claude under Customize > Skills, enable it, test whether it triggers, then use it after Cowork gathers the month's evidence. The Claude Skills tool hub is useful further reading, but this page includes enough setup detail to run the workflow.

Problem and who this is for

A monthly client recap often fails in one of two ways. It is either visually consistent but thin on substance, or it has useful evidence but takes too long to reshape into the same client-safe format every month.

This workflow is for freelancers, consultants, and small agencies who send recurring monthly recaps to clients. The hybrid setup is deliberate: Cowork gathers evidence from the local client workspace, a private monthly-client-recap Skill shapes the recap, and the official brand-guidelines Skill is optional style polish at the end.

Why Cowork fits this workflow

Monthly client recap evidence usually lives across local deliverables, screenshots, notes, metric exports, source files, and short context files. The hard part is not only writing the recap. It is finding the right evidence, excluding unrelated or old material, connecting each claim to a source, and flagging anything that still needs support before the recap becomes client-facing.

Cowork fits this workflow because it can work inside the scoped local client workspace you choose in Claude Desktop. Use it as the evidence-gathering layer: inspect the current month's source material, prepare a source-mapped evidence summary, identify unsupported claims, and save the evidence summary beside the final recap for review.

Why not normal chat or just a Skill?

Normal Claude chat can draft a client recap if you manually paste all evidence into the conversation. That can work for a one-off recap, but it adds friction every month because you still have to gather files, choose screenshots, copy metrics, exclude old source material, and trace claims back to their sources yourself.

A Skill keeps the recurring recap structure stable, but it should run after Cowork prepares the evidence summary. The private monthly-client-recap Skill is the reusable recap logic layer: sections, evidence rules, blocker language, next priorities, open decisions, and unsupported-claim handling. The brand-guidelines Skill is optional final polish after the facts are approved.

| Tool | Role | | --- | --- | | Claude Cowork | Inspects the local client workspace, gathers evidence, maps claims to source files, and flags missing support. | | monthly-client-recap Skill | Applies the recurring recap structure, evidence rules, blocker language, next-priority sections, and open-decision handling. | | brand-guidelines Skill | Optional final polish for voice, tone, formatting, naming, and presentation after facts are approved. | | Normal Claude chat | Useful for a one-off recap only if the user manually pastes all source evidence into the chat. |

What skill you will use in this workflow

You will build a private custom Skill called monthly-client-recap. The Skill tells Claude how to turn a prepared evidence summary into a monthly client recap with the same sections each time: work completed, outcomes or results, open items or blockers, next priorities, and decisions that still need client input.

This job is repetitive enough to deserve a Skill because the recap structure should not change every month. The Skill can hold the section order, the rule against invented metrics, and the difference between the recap's core logic and optional brand polish. Cowork project instructions keep the workspace rules reusable. The Skill keeps the recap logic reusable. They are complementary, not redundant.

Prerequisites

• Claude Desktop with Cowork enabled for the local workspace you choose.
• Claude with Skills available and code execution enabled.
• One local client workspace with the month's deliverables, screenshots, notes, and metrics.
• A private monthly-client-recap Skill built from the starter below.
• Optional brand rules for voice, naming, formatting, and presentation polish.
• A human review pass for metrics, screenshots, and client-facing claims.

Set up this workflow as a Cowork project

Create one Cowork project per recurring client recap, or one month-end recap project if you use the same reporting pattern across similar clients. In Claude Desktop, open Cowork, create a project from scratch or from an existing local folder, and keep the project scoped to the client material needed for the recap.

Use this folder structure:

source-material/
evidence-summary/
final-recaps/
archive-prior-months/
recap-source-notes.txt
brand-rules.txt

Use source-material for the current month's deliverables, screenshots, metric exports, notes, and source files. Use evidence-summary for Cowork's prepared source map, missing-support list, and open decisions. Use final-recaps for approved drafts or sent recaps. Move older months into archive-prior-months so Cowork does not treat old evidence as current unless you explicitly ask for comparison.

Use recap-source-notes.txt for context that is not obvious from the files, such as launch dates, approved language, known client sensitivities, or which metric export is the source of truth. Use brand-rules.txt only for voice, naming, formatting, and presentation rules.

Add reusable project instructions so you do not paste the same recap rules every month:

{
  "project_name": "Monthly Client Recap",
  "project_purpose": "Use Claude Cowork to inspect a scoped client workspace, prepare a source-mapped evidence summary, and then use a monthly-client-recap Skill to draft a consistent client-safe recap.",
  "project_instructions": [
    "Use only the current month's source-material folder unless I explicitly approve another source.",
    "Create a source-mapped evidence summary before drafting the recap.",
    "Separate verified metrics, qualitative wins, screenshots, deliverables, blockers, open decisions, and unsupported claims.",
    "Do not invent metrics, results, delivery dates, or promises.",
    "Keep screenshots tied to the specific claim they support.",
    "Apply the monthly-client-recap Skill only after the evidence summary is reviewed.",
    "Apply brand-guidelines only after the factual recap is approved.",
    "Save final recaps in final-recaps using the client name and month."
  ],
  "file_naming": "client-monthly-recap-client-name-yyyy-mm"
}

These project instructions govern how Cowork handles the workspace. The private Skill governs how the reviewed evidence becomes a recap.

How to gather the source material

1. Create one local month-end recap folder for the client.
2. Add the source material Cowork should inspect: metrics exports, screenshots, PDFs, final deliverables, status notes, project management exports, and short context notes.
3. Add one file called recap-source-notes.txt with any details that are not obvious from the files, such as launch dates, approved wording, or known client sensitivities.
4. Add one optional file called brand-rules.txt with voice, naming, formatting, and presentation rules. Keep this separate from the recap logic.
5. Remove credentials, unrelated client history, internal margin notes, and files you do not want surfaced in the recap.
6. Label screenshots and metrics clearly enough that a claim can be traced back to the right source.

Build the custom skill for this workflow

Create a folder named monthly-client-recap. Inside it, create a file named Skill.md. Paste the starter content below into that file, then adjust the sections, examples, and guardrails for your own monthly reporting process.

After the folder is ready, zip the monthly-client-recap folder itself so the ZIP contains the skill folder at its root. Do not zip loose files by themselves. In Claude, go to Customize > Skills, upload the ZIP file, and enable the Skill. Then run the trigger test below before using it on a real monthly recap.

---
name: monthly-client-recap
description: Use a reusable monthly client recap skill after Claude Cowork gathers deliverables, notes, screenshots, and metrics, then optionally apply brand-guidelines for presentation polish.
---

# Monthly Client Recap

Use this skill after Cowork or another preparation step has already gathered monthly client evidence such as notes, deliverables, screenshots, and metrics.

## Required sections
- Work completed
- Outcomes or results
- Open items or blockers
- Next priorities

## Guardrails
- Do not invent metrics or outcomes.
- Keep screenshots or examples tied to the right claim.
- Treat brand-guidelines as optional final polish, not the core logic.
- Keep the tone client-safe and concrete.

## Examples
- Use this month's gathered material to draft a monthly client recap.
- Apply my monthly client recap skill after Cowork collects the evidence.
- Draft a client recap and keep unsupported claims out.

Treat this as a starter template, not a perfect final Skill. Add your preferred headings, client-specific boundaries, screenshot rules, decision language, and any details that help the recap stay useful without overclaiming.

Optional official Anthropic skill

Anthropic's official brand-guidelines Skill is a good shortcut when the recap already has the right facts and structure, but needs presentation polish against a brand guide. Use it for voice, tone, naming, formatting, and style consistency.

The private monthly-client-recap Skill is still the better fit for repeatable recap logic. It can encode your fixed sections, evidence rules, screenshot handling, blocker language, and next-month priority structure. Treat brand-guidelines as a style layer after the recap is factually sound, not as the Skill that decides what belongs in the recap.

Skill-creator helper

Anthropic's official skill-creator Skill can help draft or refine the private monthly-client-recap Skill. Use it to improve the Skill description, trigger examples, and guardrails, then keep your client-specific reporting rules in the private Skill.

Why this should be a custom skill, not a one-off prompt

A monthly recap is recurring client operations work. A Skill keeps the same sections, evidence standards, blocker wording, and next-priority structure available every month. That reduces prompt rewriting, keeps the output more consistent, and makes it harder to forget a required review step or include an unsupported claim.

The Skill does not replace Cowork. Cowork handles the local workspace and source preparation. The Skill handles the recap rules after the evidence summary is reviewed. That order matters because style and structure are only useful when the underlying claims are supported.

Skill trigger test

After enabling the Skill, test it with a small sanitized evidence set before using it on a live client recap. These prompts should cause Claude to use the monthly-client-recap Skill:

• Apply my monthly client recap skill to this prepared client-evidence set.
• Use the monthly recap workflow after Cowork gathers the month's notes and metrics.
• Draft a client recap from this material and keep unsupported claims out.
• Turn this evidence summary into a client recap with work completed, outcomes, blockers, and next priorities.

If Claude does not appear to use the Skill, tighten the Skill description and examples with phrases like "monthly client recap," "Cowork gathers evidence," "work completed," "outcomes," and "unsupported claims."

Numbered workflow steps

1. Gather and clean the source material

Put the client files, screenshots, metrics, deliverables, and recap-source-notes.txt in one working folder. Remove unrelated files and label sources clearly before you let Cowork inspect them.

Use Cowork to prepare a structured evidence summary:

{
  "task": "Collect the month's client-delivery evidence for a monthly recap",
  "input": {
    "client_workspace_folder": "LOCAL FOLDER PATH",
    "source_notes_file": "recap-source-notes.txt"
  },
  "rules": [
    "Inspect the local files and summarize what was actually delivered or achieved this month.",
    "Separate verified metrics from qualitative wins.",
    "Tie screenshots, examples, and metrics to the source file they came from.",
    "Flag missing numbers, unsupported claims, and conflicting evidence.",
    "Prepare one clean evidence summary for the monthly recap Skill."
  ],
  "output": {
    "evidence_summary": "Structured source-based summary",
    "source_map": "Where each claim, metric, or screenshot came from",
    "missing_numbers_or_support": "Claims that still need proof",
    "open_decisions": "Items that need client or internal confirmation"
  }
}

Save Cowork's output in evidence-summary so the source map, missing-support list, and open decisions stay visible during review.

2. Review the evidence summary

Before using the Skill, check the evidence summary manually. Confirm each metric, screenshot, deliverable, blocker, and open decision is tied to the right source. Remove or rewrite any claim that the source material does not support.

Do not move to the recap draft until the evidence summary is accurate enough for client-facing work.

3. Create or enable the custom Skill for this workflow

If this is your first run, build the monthly-client-recap folder, add the Skill.md file, zip the folder, upload it in Customize > Skills, and enable it. If you already built it, confirm the Skill is enabled before you start the full monthly run.

The Skill should encode the recap's core sections: work completed, outcomes or results, open items or blockers, next priorities, and open decisions. Keep brand rules separate unless they are simple house rules that always apply.

4. Run a quick trigger test

Use a small sanitized evidence excerpt first. Try:

Apply my monthly client recap skill to this prepared client-evidence set. Keep unsupported claims out and show work completed, outcomes, blockers, next priorities, and open decisions.

Check that Claude uses the Skill's recap structure, flags weak claims, and does not let the brand-guidelines style layer replace the actual recap logic. If the output invents numbers, treats screenshots loosely, or skips blockers, tighten the Skill guardrails before using it on client material.

5. Run the actual workflow with the Skill active

Paste or attach the reviewed Cowork evidence summary, source map, missing-support list, and open decisions. Ask Claude to use the monthly-client-recap Skill to draft the recap from only the provided evidence.

Use a prompt like:

Use the monthly-client-recap Skill on this Cowork evidence summary, source map, missing-support list, and open decisions. Draft a monthly client recap with work completed, outcomes or results, open items or blockers, next priorities, and open decisions. Do not invent metrics, outcomes, or screenshot claims.

If you also want brand polish, apply the official brand-guidelines Skill after the factual recap draft is approved:

Now apply brand-guidelines only for tone, naming, formatting, and presentation polish. Preserve the facts, section logic, metrics, blockers, and open decisions from the reviewed recap.

6. Review the recap for accuracy and boundaries

Do a human review before sending. Check every metric against the source map, confirm screenshots support the exact claim near them, remove unsupported wins, and soften or delete wording that sounds like a guarantee.

Pay special attention to client-sensitive details: revenue, conversion rates, internal blockers, team names, future plans, and anything that might be read as a promise for next month.

7. Save, send, or reuse the result

Save the reviewed recap in the client folder, project management tool, or reporting system. If you send it by email, create a short cover note from the approved recap instead of asking Claude to summarize the raw evidence again.

After a few monthly runs, update the monthly-client-recap Skill with section wording, review rules, and examples that consistently work for your clients.

Future monthly runs

After the first setup, reuse the same Cowork project, folder structure, and monthly-client-recap Skill. Put only the current month's evidence in source-material, move prior material into archive-prior-months, and let the project instructions carry the standing rules.

Start with a short prompt:

Prepare this month's client recap from the current source-material folder.

For a stricter monthly run, use this:

{
  "task": "Prepare this month's client recap from the current source-material folder.",
  "rules": [
    "Use the project instructions.",
    "Create the evidence summary and source map first.",
    "Flag unsupported claims before drafting.",
    "After I approve the evidence summary, apply the monthly-client-recap Skill.",
    "Do not apply brand polish until the factual recap is approved."
  ]
}

The monthly reuse comes from three pieces working together: Cowork project instructions for the local workspace, the private Skill for recap logic, and the optional brand-guidelines Skill for final presentation polish. Keep the human review step between each handoff.

Tool-specific instructions

Use Cowork for the evidence-gathering part of the job. It should inspect the local workspace, summarize delivered work, map claims back to files, identify weak or missing support, and save the evidence summary in the project.

Use the page-specific monthly-client-recap Skill for the repeatable recap logic: sections, evidence standards, blocker handling, next priorities, and open decisions. This custom Skill should come before any brand polish because it defines what the recap is.

Use the official brand-guidelines Skill only after the recap is factually reviewed. It can help with tone and formatting, but it should not invent structure, metrics, results, or next steps.

Use normal Claude chat for a one-off recap only when all evidence is already pasted or attached. For a recurring client recap, normal chat does not replace the reusable local workspace, evidence summary, source map, and project instructions.

For broader setup context, see the Claude Skills hub. For building or refining the private recap Skill, use Anthropic's skill-creator as a helper.

Quality checks

1. Every metric in the final recap can be traced back to a local source in the client workspace.
2. Every screenshot or example supports the claim placed near it.
3. The recap includes work completed, outcomes or results, open items or blockers, next priorities, and open decisions.
4. Unsupported claims are removed or clearly marked for follow-up before send.
5. Brand polish improves tone and formatting without changing facts, structure, or evidence standards.
6. Sensitive information is appropriate for the client audience and does not expose unrelated internal context.

Common failure modes and fixes

The recap looks polished but says very little

Fix: strengthen the Cowork evidence summary before using the Skill. Require concrete deliverables, metrics, screenshots, and source references before drafting.

Brand voice overwhelms the facts

Fix: apply brand-guidelines only after the factual recap is approved. Keep the private monthly-client-recap Skill responsible for sections, evidence rules, and recap logic.

Numbers conflict across files

Fix: choose the source of truth yourself, correct the evidence summary, then rerun the recap Skill on the corrected summary.

Screenshots are attached to the wrong claim

Fix: require a source map from Cowork and review each screenshot manually. Do not let a screenshot imply a result unless the source supports that exact result.

The Skill does not trigger

Fix: update the Skill description and examples so they include the phrases you actually use, such as "monthly client recap," "Cowork evidence summary," "work completed," "outcomes," and "unsupported claims." Then run the trigger test again.