Internal comms Claude skill review: verdict, safety and alternatives
Internal Comms is Anthropic's Markdown skill that formats 3P updates, company newsletters and FAQ roundups. We rate it Conditional, 64/100, at Low confidence: four of its seven advertised formats have no guide.
- Verdict
- Conditional
- Score
- 64/100
- Safety grade
- A
- Confidence
- Low
- Best for
- Teams that already write weekly Progress/Plans/Problems updates or Slack newsletters and want Claude to follow that format
- Poor fit
- Incident reports, post-mortems or leadership updates that need a defined structure
- Reviewed version
- anthropics/skills commit 8a1541c (2026-09-28); folder last changed 2026-04-20 (licence only); no version
- Review date
- 2026-09-30
- Works in (verified)
- none
- Works in (claimed)
- Claude Code, Claude.ai, Claude API
- Status
- no change since review (checked 2026-10-21)
Get the skill: internal-comms folder on GitHub · SKILL.md at the reviewed commit · skills.sh listing
What is the internal comms Claude skill?
The internal comms Claude skill is Anthropic's Markdown pack for internal updates. It routes requests to four guides: 3P updates, company newsletters, FAQ roundups and a generic fallback. Four other advertised formats have no guide.
An Agent Skill is a SKILL.md package that an agent loads on demand. Our page on the different kinds of Claude skills covers the format. This one ships 6 files: a 1,511-byte SKILL.md, an Apache 2.0 LICENSE.txt and four guides. Its content dates from 2025-12-01.
| Guide file | Format it defines | Note at commit 8a1541c |
|---|---|---|
| 3p-updates.md | Weekly Progress, Plans, Problems | Publisher says "readable in 30-60 seconds" |
| company-newsletter.md | 20-25 bullets for Slack and email | Assumes a 1,000+ person company |
| faq-answers.md | Questions employees are asking | Tells Claude to flag uncertainty |
| general-comms.md | Everything else | Nine lines of generic advice |
| (no file) | Status reports, leadership updates, project updates, incident reports | Advertised in SKILL.md, no guide |
Every third-party page we checked repeats the seven-format list from SKILL.md. Only three of those formats have a dedicated guide, plus the fallback.
Is the internal comms skill worth installing?
Yes, but only for 3P updates, newsletters and FAQs. The internal comms skill scores 64/100, Conditional, at Low confidence. It is cheap and safe, yet incident reports get generic advice and nothing blocks invented metrics.
It is best for teams that already write weekly Progress/Plans/Problems updates or Slack-delivered newsletters. It is a poor fit in 2 cases:
- Incident reports, post-mortems or leadership updates that need a defined structure, since no template exists for them.
- Small companies or regulated teams that need audience and privacy checks before Slack, email and Drive content goes company-wide.
Our page on how verdict scores are set explains each measure.
| Measure | Score |
|---|---|
| Effectiveness | 66 |
| Reliability | 62 |
| Safety | 68 |
| Efficiency | 85 |
| Compatibility | 72 |
| Maintainability | 50 |
| Setup friction | 82 |
| Coverage of advertised formats | 50 |
| Factual grounding of drafts | 62 |
| Privacy and audience control | 45 |
| Fit to a company's own house style | 50 |
| Cost of adoption | 85 |
| Instruction weight vs value | 78 |
Compare it with other Claude writing skills ranked by verdict.
How many tokens does the internal comms skill use?
The internal comms skill loads 378 tokens, our measured file size per load. One guide adds roughly 150 to 800 tokens. A weekly 3P update from pasted notes costs about 3,000 input tokens by our estimate.
| What | Figure |
|---|---|
| SKILL.md, loaded each time the skill runs | 378 tokens |
| Whole folder, licence included | 5,598 tokens |
| Weekly 3P update from pasted notes, our estimate | 3,000 in / 400 out tokens |
| Same task with the skill loaded, our estimate | 4,378 in / 400 out tokens |
| Runs per week, our estimate | 2 |
Connected Slack, Drive and email searches add several thousand input tokens per newsletter or FAQ run (estimate).
Is the internal comms skill safe to install?
Yes, the internal comms skill is safe to install: it carries safety grade A. It has no code, network calls or credentials. Its guides tell Claude to read Slack, Drive and email without privacy checks.
The grade comes from reading the files at the reviewed commit, 8a1541c. Our guide to vetting what a skill reads covers the same checks. We found 8 facts in the files:
- Six files: SKILL.md, LICENSE.txt and four guides under examples/.
- No scripts, hooks, MCP config or agent manifests.
- No network endpoints or environment variables.
- Instructions to read Slack, Google Drive, email and calendar when connected.
- No guidance on privacy, audience checks or confidential material.
- No writes to disk; output is a chat draft.
- No handling for prompt injection in harvested Slack or email text.
- No content change since 2025-12-01, apart from a 2026-04-20 licence notice.
The skills.sh page shows Gen Agent Trust Hub pass, Socket pass and Snyk warn, with no reason given. Skillstore's July 2026 audit rated it low risk. It marked six static findings as false positives and flagged data aggregation without privacy guardrails.
What does the community report about the internal comms skill?
The community reports the internal comms skill as low risk but thin. A July 2026 security audit flagged data aggregation without privacy guardrails. Tessl scored its content 52%, and one meeting kit passed a smoke test with it.
| Claim | Source | Month | Signal |
|---|---|---|---|
| Seven types listed, four guides shipped | SKILL.md at 8a1541c | 2026-09 | Negative, strong |
| Progress defined as next-period work | 3p-updates.md | 2026-09 | Negative, weak |
| Newsletter assumes a 1,000+ person company | company-newsletter.md | 2026-09 | Negative, weak |
| Best practice 63%, content 52%, description 73% | Tessl quality review | 2026-09 | Neutral, strong |
| Low risk; aggregation flagged | Skillstore audit v7 | 2026-07 | Neutral, strong |
| Skill-check scored it 4/10 | dotfiles issue #484 | 2026-05 | Negative, weak |
| Team-recap step; smoke test passed | terminalsync-web PR #363 | 2026-09 | Positive, weak |
| Usable 3P layer; a private skill fits better | promptedwork.com guide | 2026-04 | Neutral, weak |
Tessl states the example files do not exist. They are present at commit 8a1541c, so we discount that part of its content score.
What are the internal comms skill's known failures?
The internal comms skill has five known failures at commit 8a1541c. Four of seven advertised formats, including incident reports, have no guide. The 3P guide misdefines Progress, and the newsletter assumes a 1,000+ person company.
- Missing guides: status reports, leadership updates, project updates and incident reports fall to general-comms.md.
- Wrong wording: the 3P guide defines Progress as work over the next period.
- Fixed house style: the newsletter hard-codes a 1,000+ person company and Slack emoji sections.
- Missing fabrication rule: the 3P guide pushes for metrics with no rule against inventing them.
- Missing privacy check: nothing confirms the audience before a company-wide send.
The worst plausible failure is a company-wide update with invented metrics or leaked private Slack content. All four third-party pages we checked list incident reports as supported. One of them says the skill can "Format incident retrospectives".
Which agents does the internal comms skill work with?
No agent is verified for the internal comms skill. Anthropic claims Claude Code, Claude.ai and the Claude API. One directory says it works with any model, while another says it needs Claude Code.
| Agent | Status | Evidence |
|---|---|---|
| Claude Code | Claimed | anthropics/skills README, example-skills plugin |
| Claude.ai | Claimed | anthropics/skills README |
| Claude API | Claimed | README pointer to the Skills API |
| Other models | Unknown | TypingMind copy page says GPT, Gemini and others work; astromvp says it "won't work with other AI agents out of the box" |
The skill is plain Markdown, and its compatibility score is 72. No run outside Claude is documented.
How do you install the internal comms skill?
Install the internal comms skill in Claude Code through the example-skills plugin from the anthropics/skills marketplace. You can copy its folder instead. In Claude.ai, it is an Anthropic example skill, with conflicting plan statements.
In Claude Code
- Add the marketplace:
/plugin marketplace add anthropics/skills - Install the bundle:
/plugin install example-skills@anthropic-agent-skills - Copy
skills/internal-comms/to~/.claude/skills/internal-comms/for a single-skill install.
The plugin installs every Anthropic example skill. Our guide on how to add skills to Claude Code covers project folders.
In Claude.ai and the Claude API
The Help Center lists skills on all plans, under Settings > Capabilities or Customize > Skills. The anthropics/skills README says the example skills are available to paid plans. API users upload the folder as a custom skill through the Skills API. Our guide to skills in the Claude app walks through the settings.
How do you use the internal comms skill for team updates?
Use the internal comms skill by asking Claude for a 3P update, newsletter or FAQ roundup. Paste notes or connect Slack and Drive. Check every number and remove private content before sending company-wide.
Our review lists 5 tasks:
- Weekly 3P update for one team (primary).
- Company newsletter of 20-25 bullets (primary).
- FAQ roundup of employee questions (secondary).
- Meeting recap of client decisions (secondary, terminalsync-web PR #363).
- Incident or status report (claimed only, generic fallback).
Review each draft before sending:
- Verify each number against your notes or sources.
- Remove content from private channels.
- Confirm who will receive the message.
Test it yourself:
- Paste one set of team notes.
- Ask for a 3P update.
- Ask for an incident report from the same notes.
- Check every number and link in both drafts against the notes.
What are the alternatives to the internal comms skill?
The closest alternative to the internal comms skill is Anthropic's Doc Co-Authoring skill, Conditional at 67, for longer documents. Neither covers incident reports. Writing a private skill with your own house style fits best.
| Alternative | Verdict | Better choice when |
|---|---|---|
| Doc Co-Authoring (Anthropic) | Conditional, 67 | You draft a spec, proposal or decision doc in rounds |
| Brand Guidelines (Anthropic) | Conditional, 62 | You need Anthropic brand styling on output |
| ComposioHQ/awesome-claude-skills | Not evaluated | You want a list of community skills to browse |
| A private house-style skill | Not evaluated | Your team has its own 3P or newsletter rules |
Read the Doc Co-Authoring verdict for the longer-document route. You can build a house-style skill with skill-creator from your own templates.
Where is the internal comms skill on GitHub?
The internal comms skill lives in the anthropics/skills GitHub repository at skills/internal-comms. It holds SKILL.md, an Apache 2.0 licence and four guides. GitHub code search finds 2,328 SKILL.md files with the same name.
- Repository: anthropics/skills, 15 contributors.
- Reviewed commit: 8a1541c, dated 2026-09-28.
- Last folder change: b9e19e6 on 2026-04-20, a licence notice.
- Files: 6.
- Installs on skills.sh: 75.4K.
- Same-name SKILL.md files in code search, forks included: 2,328.
Other publishers, such as davekilleen and AutumnsGrove, host copies. Check that a copy matches the anthropics/skills source before you install it.
Common questions
What skills do I need to use Claude Code?
None are required to start. Skills add repeatable workflows, and our Claude Code skills setup guide shows how to load them.
How do I create my own internal comms skill?
Use Anthropic's skill-creator skill with your team's 3P and newsletter templates. It turns those templates into a SKILL.md package for your team.
Which Anthropic skills make documents?
Anthropic's PDF, PowerPoint and Excel skills make office files, and its DOCX skill handles Word files. See our PDF skill review, PowerPoint skill review and Excel skill review.