Download Critical Compass

What is Critical Compass?

Critical Compass is a critical-thinking toolbox for people who need written work to hold up under scrutiny. Paste in a grant narrative, advocacy draft, public statement, report, or article, and it helps spot weak evidence, hidden assumptions, framing gaps, missing voices, and claims to verify before you publish or submit. It can be used through the plugin, MCP/extension, or skill files below.

Claude Cowork Plugin

Best for Claude Cowork or Claude Code. This bundle includes the Critical Compass MCP/extension, skill, and predefined agents.

v0.1.0-beta.11

Download Plugin

→ How to install

Claude Desktop Extension

Best if you use Claude Desktop. This is the MCP connector installed through Settings → Extensions.

v0.1.0-beta.11

Download Extension

→ How to install

Skill / Prompt-Only Alternative

A standalone skill Claude can read without the MCP connector. Useful on its own or alongside the plugin or extension.

v0.1.0-beta.11

Download Skill Download Skill Zip

→ How to install

Pick Your Path

Downloads are live for v0.1.0-beta.11. Choose the plugin for Claude Cowork or Claude Code, the MCPB/Desktop Extension for Claude Desktop, or the standalone skill when tool access is unavailable.

I want a fast first read	Use the quick audit prompt for a plain-language risk read, the top concerns, and one next action before you share or cite a text.
I am teaching a class	Use the classroom bias-audit prompt so students supply the same audience, intended use, text type, audit depth, and output section order.
I need the full audit	Use the full standard audit prompt for the Critical Integrity Snapshot, evidence gaps, reasoning traps, assumptions, missing voices, and what holds up.
I want questions and calibration	Use the advanced human-loop prompt when context, sensitivity, stakeholder perspective, or report readiness needs user input before the audit is finalized.
I need a shareable deliverable	Use the report-ready workflow after an audit payload has been validated, then generate Markdown, HTML, or PDF artifacts.
I only want the prompt layer	Install the skill or skill zip when the MCP connector is unavailable, or keep it alongside the plugin/extension as a lightweight fallback.

You can install more than one. Start with the plugin, extension, or skill that matches your Claude surface, then use the path above to decide what to do after install.

Want your AI assistant to help install this? Use the AI-assisted install prompt. Using Codex? See the FAQ tab for technical-user guidance.

Your First Prompt

Once Critical Compass is installed, copy one of these prompts into your AI to get started right away.

Skill Audit

When to use: You have the Critical Compass skill installed and want a full audit without the MCP.

I have the Critical Compass skill installed.

Before you begin the audit, ask me the following questions one at a time
so you can calibrate your analysis:

1. What type of content is this? (e.g., article, social post, study,
   speech, email, memo, draft I wrote)
2. What domain does it fall under? (e.g., health, politics, AI, business,
   education, general)
3. What was your relationship to this text - did you write it, read it,
   or receive it?
4. What's your take going in - did anything already feel off, biased,
   or unsupported when you read it?
5. What kind of feedback are you looking for - do you want to improve it,
   decide whether to trust it, share it, or act on it?

After I answer, run a full Critical Compass audit using the skill. Return:
- Critical Integrity Snapshot (plain-language read + five-dimension table)
- Before Using This Text (one practical caution before I act on, share,
  or publish this)
- Decision Read: Trust, Revise, Verify, or Reframe - one action with rationale
- Top concerns (bias, fallacy, assumptions, evidence gaps) with confidence levels
- What holds up
- What would change your mind
- 3 follow-up questions and 3 next steps

Text to audit:
"""
[Paste the text, claim, article, or argument here.]
"""

Quick Audit with MCP

When to use: You have the MCP connector or extension active and want a faster, knowledge-base-backed audit.

I have the Critical Compass MCP connector/extension active.

Start by calling critical_compass_overview to orient yourself on the available
tools and confirm the right path for a standard audit. You can also use
quick_scan as part of your orientation if helpful.

Then ask me these questions before running the audit:
- What type of content is this?
- What's the domain? (health / politics / AI / business / education / general)
- Why are you auditing it - trust check, deciding to share, fact-checking,
  other?
- Anything you already noticed that felt off?

After I answer, run audit_text on the content below. Return:
- Critical Integrity Snapshot (plain-language read + five-dimension table)
- Before Using This Text
- Decision Read: one primary action (Trust / Revise / Verify / Reframe)
  with one rationale sentence and one concrete next move
- Top 3-5 concerns with match confidence and source confidence
- What holds up
- 3 follow-up questions and 3 next steps

Content to audit:
"""
[Paste the text here.]
"""

Classroom Bias Audit

When to use: You are using Critical Compass in a course or assignment and need stable section order for peer discussion.

Please run a standard Critical Compass audit on the text below.

Use this fixed frame:
- Audience: [who this is for]
- Intended use: [what this writing is meant to do]
- Text type: [memo / pitch / email / report section / social post / proposal / other]
- Audit depth: standard
- Output purpose: classroom bias-audit reflection

Do not run an advanced panel review, multi-agent review, or web research. Do not rewrite the text unless I ask afterward.

Use this exact output structure:
1. Critical Integrity Snapshot
2. Plain-Language Read
3. Before Using This Text
4. Top 3-5 findings
   - finding name
   - severity: high / medium / low
   - specific evidence from the text
   - why it matters for the stated audience and intended use
5. Key tension
6. Recommended next revision
7. Two peer-discussion questions

Here is the text to audit:

[paste text]

Full Standard Audit with MCP

When to use: You want a full structured audit with the MCP connector or extension, but not the multi-agent pipeline.

I have the Critical Compass MCP connector/extension active.

Call critical_compass_overview first to confirm the right audit path and
available tools for a full standard audit.

Then ask me the following before running anything:
- Content type: (article / study / policy doc / memo / email / post / other)
- Domain: (health / politics / AI / business / research / education / general)
- Your relationship to the text: did you write it, read it, or receive it?
- Anything you already noticed that felt off, biased, or unsupported?
- Stakes: what are you deciding - publish it, share it, act on it, cite it?
- Sensitivity flag: does this touch health, identity, politics, religion,
  or legal topics? (yes / no / unsure)

After I answer, run audit_text on the content below. Return the full audit:
- Critical Integrity Snapshot (plain-language read + five-dimension table)
- Before Using This Text
- Decision Read: Trust / Revise / Verify / Reframe with rationale and
  one concrete next move
- Top concerns with match confidence and source confidence
- Argument map (if a central claim is present)
- Assumptions (testable and not testable)
- Evidence Needed Next
- Reasoning Trap Checks (if visible cues exist)
- Alternative frames (1-2)
- What would change your mind
- 3 follow-up questions and 3 next steps

Content to audit:
"""
[Paste the full content here.]
"""

Advanced Audit with Multi-Agent Support

When to use: You want the full advanced pipeline with multi-agent support, human-in-the-loop pauses, and report-ready output.

I have the Critical Compass MCP connector/extension active.

I want to run an advanced audit with multi-agent support and human-in-the-loop
review on the content below. This is the full advanced audit pipeline, not
a standard audit_text call.

Step 1 - Orient: Call critical_compass_overview to confirm tools and routing
for advanced audit mode.

Step 2 - Define agents: Call define_audit_agents for this content. Before
proceeding, show me the proposed analyst personas and pause for my approval.
I want to be in the loop before any analysis runs.

Step 3 - Independent analyses: Call run_independent_audit_analysis for each
agent. Each analyst should work independently from its distinct perspective
and use lookup_bias, challenge_claim, and get_framework as needed.

Step 4 - Map tensions: Call map_audit_tensions to surface where agents
disagree, where evidence is thin, and which findings could change the verdict.

Step 5 - Structured debate: Call run_audit_debate on the highest-tension finding.

Step 6 - Synthesize: Call synthesize_audit_verdict to produce a final verdict,
confirm or revise the Critical Integrity Snapshot, and identify what holds
across agents versus where genuine disagreement remains.

Step 7 - Human-in-the-loop: Pause before report generation. Surface the
human-loop questions from the audit pipeline and wait for my responses
before proceeding.

Step 8 - Report: After human-in-the-loop is complete, use the
report_ready_payload from synthesize_audit_verdict, run
validate_report_payload first, and only then call generate_audit_report
with the validated audit data.

The primary report output is a PDF. Tell me when the PDF is ready and provide
the file path.

If I also want a Markdown version, generate that alongside the PDF.

If I want a browser-friendly version too, provide the generated viewer
path or call generate_audit_artifact_viewer if a separate viewer is needed.

Pause at every step where my judgment is needed. Do not proceed past a
human-in-the-loop gate without my input.

Content to audit:
"""
[Paste the full content here.]
"""

Context:
- Content type: [article / study / policy doc / speech / report / other]
- Domain: [health / politics / AI / business / research / other]
- Why this matters: [e.g., about to publish / present / make a decision based on it]
- Preferred agent types (optional): [leave blank to let the MCP select, or name lenses - e.g., policy analyst, methodologist, steelman advocate, community voice, missing-perspective reviewer]
- Desired report outputs: [PDF only / PDF + Markdown / PDF + Markdown + viewer]

Coaching Mode (Skill Only)

When to use: You wrote the text and want help strengthening it without running a full audit.

I have the Critical Compass skill installed.

I want coaching mode - not an audit of someone else's writing. I wrote
this and I want help making it stronger, fairer, and more credible to a
skeptical reader. Don't tear it apart - calibrate it.

Before you start, ask me:
1. Who is my intended audience?
2. What am I trying to accomplish with this piece?
3. What am I most uncertain or worried about in it?

After I answer, use the Critical Compass coaching rewrite mode. Give me:
- What holds up and should stay as-is
- What's overclaimed or unsupported (and why)
- A stronger version (clearer, more persuasive without overclaiming)
- A more evidence-calibrated version (preserves my core point, signals
  what's known vs. uncertain)
- The single most important revision to make first

What I wrote:
"""
[Paste your draft, argument, post, or claim here.]
"""

My goal with this writing:
"""
[e.g., I'm trying to persuade my team / submit this as a proposal / post this publicly / use this in a report.]
"""

About the author

Jonathan Kyle Hobson is a UX researcher, product strategist, and AI tool builder with 10+ years of experience across design, storytelling, and emerging technology. He holds a master's degree in User Experience from Arizona State University. Kyle created Critical Compass to give nonprofits, advocacy teams, and public communicators a practical, rigorous way to pressure-test their work, because high-stakes writing deserves the same scrutiny as high-stakes decisions.

Connect on LinkedIn

What Is Critical Compass?

Critical Compass is a local critical-thinking assistant for pressure-testing written work before it is published, submitted, shared, or used to make a decision. It is designed for drafts that need to hold up under scrutiny: grant narratives, advocacy writing, public statements, reports, articles, policy arguments, and other high-stakes text.

It does not just proofread. Critical Compass looks for the reasoning underneath the writing: the central claims, evidence gaps, assumptions, possible bias patterns, missing stakeholder perspectives, framing risks, audience fit, and claims that need verification. It helps answer a practical question: if someone smart and skeptical read this, where would it hold up and where would it break?

For non-technical users, it works like a structured review partner. For technical users, it exposes a local audit workflow with tools for quick scans, full audits, claim stress tests, reasoning-framework lookup, saved audit results, report validation, report generation, and advanced multi-perspective review.

How It's Built

Critical Compass is distributed as a local MCP/extension, a Claude Cowork or Claude Code plugin, and a prompt-only skill. The MCP exposes tools for onboarding, routing, quick scans, full audits, advanced audits, claim stress tests, fact-check planning, knowledge-base lookup, saved results, pattern tracking, report validation, and report generation.

Under the hood, it combines structured audit workflows with a critical-thinking knowledge base. That knowledge base includes biases, fallacies, reasoning frameworks, thinking lenses, methodology checks, reasoning flows, and reusable audit scaffolds. The workflow layer turns those resources into repeatable review steps instead of asking the model to invent an audit process from scratch.

The system is local-first and scaffold-first by default. It can help identify what should be verified, but it does not require Google, Brave, provider keys, or live retrieval for its core audit workflows. Advanced audit paths add role and persona scaffolds for multi-perspective review while keeping the user in control of the final judgment.

What's Inside

A compact map of the major MCP surfaces — what they do and how to ask for them.

get_started

First-session onboarding: clarifies goal, intended audience, and audit depth, then routes to the right tool.

Say: “I'm new - help me get started with Critical Compass.”

critical_compass_overview

Friendly map of workflows, routing rules, case examples, and starter prompts.

Say: “Give me an overview of Critical Compass.”

quick_scan

Lightweight gut-check with a Critical Integrity Snapshot, top risks, key tension, and one next action.

Say: “Quick-scan this text before I decide whether it needs a full audit.”

audit_text

Full audit for assumptions, bias patterns, evidence gaps, argument structure, check-worthy claims, alternative frames, and audience fit.

Say: “Run a full Critical Compass audit on this text.”

advanced_audit

High-stakes multi-perspective review with analyst roles, tension mapping, debate, dissent, and synthesis.

Say: “Run an advanced audit on this draft.”

lookup_bias / get_framework / search_knowledge_base

Knowledge-base lookup tools for verified bias definitions, thinking lenses, frameworks, and reasoning flows.

Say: “Search Critical Compass for this concept before citing it.”

save_audit_result / compare_audit_results

Saved-result tools for longitudinal review, score comparison, bias changes, and framing-drift scaffolding.

Say: “Save this audit result, then compare it to the revised version.”

validate_report_payload

Read-only dry run for report schema, field constraints, renderer diagnostics, and preflight blockers before rendering.

Say: “Validate this report payload before generating the PDF.”

generate_audit_report

Turns completed audit data into a stakeholder-ready PDF, Markdown file, manifest, and local viewer.

Say: “Generate a PDF report from this completed audit.”

How I Use It

I use Critical Compass when I want to slow down before trusting, publishing, citing, sharing, or acting on a piece of writing. It is especially useful for articles, research papers, drafts, or other writing someone sends me when I need to know whether the evidence, assumptions, framing, and audience fit are actually strong enough.

For quick reads, I use a quick scan for the plain-language read, the Before Using This Text caution, and the one action worth taking next. For important drafts, I use a full audit so the tool can map claims, surface evidence gaps, flag reasoning traps, and suggest what would make the argument more credible. For high-stakes or politically sensitive work, I use advanced audit mode because multiple perspectives are better than one confident pass.

I use it as a decision-support tool, not as an automatic judge. Claude is strong at conversational synthesis and stakeholder-sensitive framing; Codex is useful when I want the audit turned into structured artifacts, reports, or implementation-ready follow-up work. In both cases, Critical Compass gives the model a disciplined review path instead of leaving the critique up to vibes.

See examples →

Install Guide

Choose the install path for the Claude surface you use. These screenshots show the tested UI. If your app does not show this path, stop and use the diagnostic prompt instead of improvising.

Tested Claude Surfaces

The screenshots below reflect the Claude Code, Claude Cowork, and Claude Desktop surfaces available during testing. Claude's install UI can differ by platform, account, and version. Local MCP/plugin and Desktop extension installs are not claimed for Claude.ai in a web browser.

Before Manual Config Edits

Do not hand-edit Claude config, or let an assistant rewrite it, unless it first backs up the file, reads the existing server names, merges only critical-compass, writes UTF-8 without a BOM, and verifies that no existing MCPs disappeared.

Claude Cowork Plugin

Download critical-compass-plugin.zip.
Open Claude Cowork or Claude Code.
Click Customize in the left sidebar.
Under Personal Plugins, click the + button.
Hover over Create Plugin, then click Upload Plugin.
Drag the zip file into the upload area.
Use critical_compass_overview or one of the first prompts on the Get Started tab.

If you only see Skills and Connectors, and Connectors only accepts a remote MCP URL, stop. That is not the plugin upload path.

Step 1: Open Claude Code or Claude Cowork.
Step 2: Open Customize from the left sidebar.
Step 3: Confirm you can see Personal Plugins.
Step 4: Hover over the plus button until Add Plugin appears.
Step 5: Click the plus button to open the plugin menu.
Step 6: Choose Create Plugin, then Upload Plugin.
Step 7: Choose the local upload option.
Step 8: Drag in critical-compass-plugin.zip.
Step 9: Confirm Critical Compass appears as installed.

Claude Desktop Extension

Download critical-compass-0.1.0.mcpb.
Open the Claude Desktop app.
Go to Settings → Extensions.
Drag the downloaded file into the Extensions panel.
Approve the standard local-extension warning.
Ask Claude to use Critical Compass for a quick or full audit.

If your Claude Desktop build does not show Settings → Extensions, do not hand-edit config as a beginner workaround. Use the diagnostic prompt.

Step 1: Download the Critical Compass .mcpb file.
Step 2: Open Claude Desktop settings.
Step 3: Open the Extensions area if your build has it.
Step 4: Drag the .mcpb file into the Extensions panel.
Step 5: Confirm the extension is installed.

Skill File or Skill Zip

Download critical-compass.skill or skill.zip.
Open Claude or Claude Code.
Go to Skills, then click the + button.
Hover over Create Skill, then click Upload Skill.
Drag in your downloaded file.
Use this on its own, or install it alongside the plugin or extension when you want the prompt-only version available too.

Use the skill when the MCP/plugin path is unavailable, or keep it as a prompt-only fallback alongside the tool connection.

Step 1: Open the Skills area.
Step 2: Click the plus button.
Step 3: Choose Create Skill, then Upload Skill.
Step 4: Select critical-compass.skill or skill.zip.
Step 5: Confirm Critical Compass is available as a skill.

After Install, Verify

Fully restart Claude if the tool list does not refresh.
Start a new chat.
Ask for get_started or critical_compass_overview.
If you had other MCPs installed before, confirm they still appear.

If the Expected UI Is Missing

Do not improvise in Connectors unless that surface clearly accepts the exact file type you downloaded. Use the skill/reference path or the diagnostic prompt instead.

Back Out Safely

Restore the latest config backup or remove only the critical-compass entry from mcpServers. Never reset the whole Claude config to remove one MCP.

If Install Fails

Use the diagnostic prompt in the Advanced tab before editing any files manually. Manual config install is a technical fallback for people who have installed MCPs this way before.

Advanced

This section is for IT reviewers, security-conscious users, and anyone troubleshooting installation.

View full release notes

Download source files

Use this zip if you want to inspect, modify, or rebuild Critical Compass yourself. It excludes private reports, generated release files, local state, env files, caches, and machine-specific artifacts.

Download Source Zip

Advanced: Verify Download

These SHA-256 checksums help technical reviewers confirm that downloaded files match this release.

critical-compass-0.1.0.mcpb	`a9a22efd47b9b10d7b7ff603dcb763b1ac57bdd88781279602e2df49f16cc77c`
critical-compass-plugin.zip	`451bc450b61df8088a60d71ec0cb822eefb7523961a765a362e744d343e1ec3c`
critical-compass-source.zip	`effd8bf6edc4aabe5e2737ce0cbd3a4b99a1db3ae0ad7e2927919dbd86aa5bdd`
critical-compass.skill	`ba1ee07256c209e2f1a406fe4926665692c088fa896ac466c2ad68df8b842533`
skill.zip	`fe5df54382bbd097cbda82e898812f09384d986955de620b22f0e98d5c5f63c4`

AI-Assisted Install Prompt

Use this with a coding assistant that can inspect downloads, check your computer setup, and guide client-specific UI steps. Manual config edits are a technical fallback only; the assistant should preserve existing configuration and stop if the expected install surface is missing.

You are a careful, technically conservative AI install assistant. Your job is to help me add or install Critical Compass - optionally including the plugin, MCPB/Desktop extension, skill, or local MCP source - safely, without modifying unrelated configuration or introducing unexpected dependencies. Do not delete anything; only add.

Critical Compass is a critical-thinking toolbox for pressure-testing written work before publication or submission.

Release page: https://jonathankhobson.github.io/critical-compass/

PHASE 1 - Understand my environment
Before doing anything, ask me:
1. Which AI client are you using? (Claude Cowork, Claude Code, Claude Desktop, Codex, or other)
2. Which operating system are you on? (macOS, Windows, Linux, WSL, or other)
3. Have you installed any MCPs, plugins, or skills before? (yes / no / unsure)
4. What do you want added or installed? (local MCP source / plugin / MCPB Desktop extension / skill / all)

Do not proceed to Phase 2 until I answer all four questions.

PHASE 2 - Identify the right install path
Based on my answers, recommend the appropriate install path(s). Multiple paths can be installed together.

Install methods and what you can automate:
- Plugin zip - add/install support depends on the client. Download and inspect the plugin zip, then determine whether the current client supports adding it automatically or whether I need to complete a UI step.
- MCPB/Desktop extension file - add/install support depends on the client. Download and inspect the MCPB file, then determine whether the current client supports adding it automatically or whether I need to complete a UI step.
- Skill or skill zip - add/install support depends on the client. Download and inspect the skill file, then determine whether the current client supports adding it automatically or whether I need to complete a UI step.
- Local MCP source zip - technical fallback only. Use it only if I have installed MCPs this way before or I explicitly confirm I want a config-based install after the UI paths fail.

Important distinction:
- Treat plugin, MCPB/Desktop extension, and skill upload as the preferred user paths when the visible UI supports them.
- For plugin, MCPB/Desktop extension, and skill files, do not assume the add/install method. Check the current client, current docs, local files, and visible UI. Do as much safely as you can, then guide me through any step that requires my action.
- Treat local MCP source and manual config edits as a technical fallback, not the beginner path.
- If the UI or install method has changed, use the current official docs or the client's visible UI rather than stale instructions.

For each path you recommend, state in one sentence what you can do automatically and what I may need to do manually. Ask me to confirm before proceeding.

PHASE 3 - Review before installing
Open the release page and identify the current version.
Download only the file(s) needed for the confirmed install path(s).
Before installing, check for:
- Unexpected network calls or external API dependencies
- Hidden API keys or credential requirements
- Suspicious scripts or broad file system access
- Unrelated services bundled in

If you find anything concerning, STOP. Describe the concern clearly and ask me how to proceed. Do not install until I confirm.

PHASE 4 - Install safely
For a local MCP source install:
- Prefer a stable local install folder and avoid running directly from a temporary download folder.
- Read existing config files before editing them.
- Preserve all unrelated MCP servers and settings.
- Back up any config file before changing it.
- Prefer a direct interpreter command with absolute paths instead of shell wrappers.
- If writing Claude Desktop JSON on Windows, write UTF-8 without a byte-order mark. Do not use Windows PowerShell 5.x Set-Content -Encoding UTF8 for the final config write.
- After writing, verify the first three bytes are not EF BB BF and verify every pre-existing mcpServers entry is still present.

Common config locations to check, then verify against current docs and the actual client:
- Claude Desktop on macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
- Claude Desktop on Windows: %APPDATA%\Claude\claude_desktop_config.json
- Claude Desktop on Linux or unofficial builds: location may vary; check current docs and the app config directory before editing.
- Codex on macOS/Linux: ~/.codex/config.toml
- Codex on Windows: %USERPROFILE%\.codex\config.toml
- Claude Code project MCP config: .mcp.json in the workspace root

Known local-MCP troubleshooting rules:
- Existing Codex sessions usually do not hot-load new MCP config; start a fresh session after editing.
- Claude Desktop usually needs a full quit/reopen after config changes.
- For Codex, validate the config if a validator is available before restarting.
- For Claude Code, confirm the user opened the workspace root that contains .mcp.json.
- If files live on /Volumes or another external drive, avoid shell wrappers when possible and prefer a direct interpreter command.
- If Python behaves strangely on macOS external drives, check for AppleDouble sidecars such as ._* or .__* and consider keeping the runtime in a stable home-directory path while pointing it at the project/data.

For plugin, MCPB/Desktop extension, or skill installs:
- Do not hand-edit installed plugin, extension, or skill cache folders.
- Download and inspect the package, then determine the current add/install method for the user's client.
- If a UI step is required, guide me through it clearly and wait for me to confirm completion.
- If the expected UI path is missing, stop. Do not improvise in Connectors unless that surface clearly accepts the exact file type being installed.
- Do not add paid services, external APIs, provider keys, or unrelated servers unless I explicitly ask.

PHASE 5 - Smoke test
After installing, confirm Critical Compass is available:
- If you installed the local MCP source on Claude Desktop, Claude Code, or Codex, remind me to restart or start a fresh session before expecting the MCP to appear.
- If MCP tools are active: call get_started or critical_compass_overview.
- If I had other MCPs before install, confirm they are still present. If any disappeared, stop and restore from the pre-install backup.
- If only the skill is available: ask Critical Compass to pressure-test this sentence: "Our program has demonstrated significant impact across diverse communities."
- If install fails: tell me whether the issue is package validation, MCP launch, config syntax, missing dependency, wrong workspace root, client restart needed, or unsupported-client behavior - and suggest one concrete next step.

Constraints throughout:
- One phase at a time. Do not skip ahead.
- Ask rather than assume when uncertain.
- Prefer the most conservative option when two paths are equally valid.
- Preserve unrelated configuration. Do not delete anything; only add.
- Use current docs or visible client UI when the install method is unclear.

Diagnostics Prompt

If installation fails, copy this prompt into Claude, Claude Code, or Codex and attach the file you tried to install.

# Critical Compass MCPB Installer / Diagnostic Prompt

Paste this into Claude, Claude Code, Codex, or another capable coding assistant when you want help installing or diagnosing Critical Compass.

```text
You are helping me install or diagnose the Critical Compass MCPB/Desktop Extension in Claude Desktop.

Goal:
- Get Critical Compass installed and visible as an MCP server/tool set.
- Do not delete or overwrite any existing Claude Desktop MCP configuration.
- Preserve unrelated MCP servers and settings.
- Prefer MCPB/Desktop Extension installation over manual JSON editing.
- Treat manual config editing as a technical fallback only for users who have installed MCPs this way before.

Context:
- The package file should be named like critical-compass-0.1.0.mcpb.
- Critical Compass is a local MCP server for pressure-testing text before use.
- It should expose tools such as get_started, critical_compass_overview, quick_scan, audit_text, claim_stress_test, factcheck_lookup, and generate_audit_report.
- factcheck_lookup is scaffold-first by default; it does not require Google, Brave, API keys, or live retrieval.

Steps:
1. Confirm Claude Desktop supports Extensions or MCPB/Desktop Extensions.
2. Help me install the .mcpb through Claude Desktop Settings > Extensions.
3. Restart Claude Desktop if needed.
4. Check whether Critical Compass appears in the available MCP tools.
5. Run a small smoke prompt:
   "Call get_started for Critical Compass and tell me the first five-minute path."
6. If it fails, diagnose in this order:
   - Is the .mcpb file present and not still zipped inside another archive?
   - Is Claude Desktop updated enough to show Settings > Extensions?
   - Did UV dependency setup fail on first launch?
   - Are there Claude Desktop MCP logs showing a process launch error?
   - Are writable files being redirected to user app data rather than the extension folder?
   - Is any failure specific to PDF rendering or OCR rather than the core audit tools?
   - If manual config editing is being considered, did you back up the file, read existing `mcpServers`, and confirm the user is comfortable with this technical fallback?

Safety rules:
- Do not add Google/Brave/provider keys unless I explicitly ask.
- Do not claim factcheck_lookup performed retrieval unless lookup_performed=true.
- Do not treat "no match" as true or false.
- Do not change unrelated servers while diagnosing this install.
- If manual config inspection is needed, read the current config first and preserve all existing entries.
- If manual config writing is needed, merge only the `critical-compass` entry into `mcpServers`; do not rewrite or shrink the whole config.
- On Windows PowerShell 5.x, do not write Claude config with `Set-Content -Encoding UTF8`; it can add a UTF-8 BOM. Use a no-BOM writer and verify the file does not start with `EF BB BF`.
- After any config write, print the before/after MCP server names and confirm all pre-existing servers are still present.

Expected result:
- Critical Compass is installed.
- get_started and critical_compass_overview work.
- Any pre-existing MCP servers still appear after restart.
- If a feature is unavailable, explain whether it is an install issue, dependency issue, renderer/OCR limitation, or normal scaffold-first behavior.
```

Alternative Share Link

A Claude Artifact mirror is available as a secondary way to preview this page. Downloads still point to the GitHub release files.

View Claude Artifact mirror

What is Critical Compass?

Claude Cowork Plugin

Claude Desktop Extension

Skill / Prompt-Only Alternative

Pick Your Path

Your First Prompt

Skill Audit

Quick Audit with MCP

Classroom Bias Audit

Full Standard Audit with MCP

Advanced Audit with Multi-Agent Support

Coaching Mode (Skill Only)

About the author

What Is Critical Compass?

How It's Built

What's Inside

How I Use It

Example

Finished Audit Snapshot

Claude Conversation Snapshot

Install Guide

Tested Claude Surfaces

Before Manual Config Edits

Claude Cowork Plugin

Claude Desktop Extension

Skill File or Skill Zip

After Install, Verify

If the Expected UI Is Missing

Back Out Safely

If Install Fails

Frequently Asked Questions

What does Critical Compass actually do?

Who is it for?

What is an MCP connector or extension?

What is the difference between the plugin, extension, and skill?

Does this work with Codex?

Is Critical Compass safe to install?

Do I need a GitHub account to download?

Advanced

Download source files

AI-Assisted Install Prompt

Diagnostics Prompt

Alternative Share Link