Skip to content

Ravi CI Guard: GitHub Actions security audit MCP server

dev.workers.agent-arena-base.ravi-ci-guard/ci-guardravi-ci-guard.agent-arena-base.workers.devstreamable-httpv1.1.0Scanned

Audit GitHub Actions workflows: pwn requests, script injection, token scope, SHA pins, fix patches

7 tools: 2 do something that can't be undone, 4 change things and 1 only read.

Tools: 0 delete, 2 irreversible, 4 write, 1 read.
  • Reads private data · 0
  • Reads text others write · 5
  • Can send data out · 3
  • 4 findings

With Anlyon

2 tools here do something that can't be undone. Govern them.

  • Anlyon lists the tools and adopts none.
  • Each tool you adopt gets approvals and receipts.
  • Your agent never holds the key.
Govern this server →

A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.

Anlyon alternative

Everything your agent needs in GitHub. None of the key.

Anlyon has its own GitHub integration. Every write runs under a limit you set and comes back with a receipt read from the provider.

Anlyon Radar does not run, proxy or govern this MCP server. A tool your own code calls directly is not routed through Anlyon and is not governed by it.

See the GitHub integration →

What the Ravi CI Guard: GitHub Actions security audit MCP server can do

The Ravi CI Guard: GitHub Actions security audit MCP server is published by ravi-ci-guard.agent-arena-base.workers.dev, a domain the MCP Registry verified. It has 7 tools. It can do things that cannot be undone, such as sending, paying or publishing (payment_options and redeem_transfer_pass) and create or change data (4 tools: audit_sample, peek_workflows, audit_workflows and pin_actions), but cannot delete data. 1 tool only reads.

5 tools bring in text other people can write (audit_sample, peek_workflows, audit_workflows and 2 more); 3 tools can send data out of your system (peek_workflows, audit_workflows and pin_actions).

The fixed checks matched 4 findings (credentials-mention and annotation-conflict). Each one comes with the text it matched.

Last scanned Nov 21, 2026, version 1.1.0.

Tools

Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.

7 tools

Can't be undone 2Claude Code: ask
  • payment_optionsFREE.1 finding

    Why it is in this group

    • The name "payment_options" says "payment": a word that usually means money moves, which cannot be taken back.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    Findings

    • Warn

      The server marks "payment_options" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

      Matched text

      annotations: {"readOnlyHint":true}; The name "payment_options" says "payment": a word that usually means money moves, which cannot be taken back.

      What this check looks for
    Description, as the server wrote it
    FREE. Prices, the plain-USDC-transfer pass (0.50 / 3 / 8 USDC on Base for 24 h / 30 days + Team Kit / 90 days + Team Kit), the x402 per-call alternative, the self-hosted Team Kit, and the status of the pass you are using (if any).

    Annotations the server declares

    • readOnlyHint: true
  • redeem_transfer_passFREE.Reads text others write1 finding

    Why it is in this group

    • The name "redeem_transfer_pass" says "transfer": a word for a change that another call does not undo.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.

    Findings

    • Warn

      The server marks "redeem_transfer_pass" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

      Matched text

      annotations: {"readOnlyHint":true,"openWorldHint":true}; The name "redeem_transfer_pass" says "transfer": a word for a change that another call does not undo.

      What this check looks for
    Description, as the server wrote it
    FREE. After sending USDC on Base to the address in payment_options, give the transaction hash to get your CI Guard pass token (and the Team Kit download link for 3 USDC or more). Redeeming the same hash again returns the same token.

    Annotations the server declares

    • readOnlyHint: true
    • openWorldHint: true
Changes things 4Claude Code: ask
  • audit_sampleFREE demo.Reads text others writeNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    Description, as the server wrote it
    FREE demo. Runs the real audit_workflows on a built-in SAMPLE workflow (a pull_request_target preview job) that contains typical mistakes on purpose: pwn request, script injection, missing permissions, retired runner, unpinned and deprecated actions. Shows exactly what audit_workflows returns (findings with evidence/why/fix, SHA pins, patches) before you buy a pass. Takes no arguments; it never audits your own code.

    Annotations the server declares

    • readOnlyHint: true
    • openWorldHint: true
  • audit_workflowsSecurity + reliability audit of GitHub Actions workflows: repo=owner/name (all workflows at the default branch, or ref=), url= (one workflow file) or yaml= (the file text).Reads text others writeCan send data out1 finding

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Findings

    • Info

      The description mentions "credentials".

      Matched text

      description: "...nd comments, artifact poisoning, leaked credentials, GITHUB_TOKEN permissions, secrets: inh..."

      What this check looks for
    Description, as the server wrote it
    Security + reliability audit of GitHub Actions workflows: repo=owner/name (all workflows at the default branch, or ref=), url= (one workflow file) or yaml= (the file text). About 30 line-accurate checks: pwn requests (pull_request_target / workflow_run running PR code), script injection from PR titles, branch names and comments, artifact poisoning, leaked credentials, GITHUB_TOKEN permissions, secrets: inherit, unpinned or compromised actions, deprecated actions that now fail, retired runners, self-hosted runners on PRs, ArtiPACKED token leaks, curl|sh, set-output, invalid structure, timeouts. Returns score/grade, findings (rule, severity, file:line, evidence, why, fix), commit-SHA pins resolved live from GitHub and two git-apply patches (safe, hardening). Needs a CI Guard pass (0.50 USDC = 24 h, 3 USDC = 30 days + Team Kit, 8 USDC = 90 days + Team Kit, by plain USDC transfer on Base; see payment_options), or pay per call over HTTP x402 ($0.10) at /v1/gha/audit.

    Annotations the server declares

    • readOnlyHint: true
    • openWorldHint: true
  • peek_workflowsFREE.Reads text others writeCan send data outNo findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
    Description, as the server wrote it
    FREE. Score (0-100), grade and finding counts by severity and rule for your workflows: repo=owner/name, url= one workflow file, or yaml= pasted text. Counts only: the line-by-line findings, fixes, SHA pins and git-apply patches come from audit_workflows.

    Annotations the server declares

    • readOnlyHint: true
    • openWorldHint: true
  • pin_actionsResolve uses: references to immutable commit SHAs live from GitHub: actions=actions/checkout@v4,docker/login-action@v3 (up to 20), or repo=/url=/yaml= to pin every uses: in those w…Reads text others writeCan send data out1 finding

    Why it is in this group

    • The name "pin_actions" says "pin": a word for a change.
    • The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.

    What it can reach

    • untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
    • outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.

    Findings

    • Warn

      The server marks "pin_actions" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

      Matched text

      annotations: {"readOnlyHint":true,"openWorldHint":true}; The name "pin_actions" says "pin": a word for a change.

      What this check looks for
    Description, as the server wrote it
    Resolve uses: references to immutable commit SHAs live from GitHub: actions=actions/checkout@v4,docker/login-action@v3 (up to 20), or repo=/url=/yaml= to pin every uses: in those workflows. Per action: SHA (annotated tags peeled), the exact version tag for the "# v4.2.2" comment, a ready uses: line, latest release and whether a newer major exists. Needs a CI Guard pass (0.50 USDC = 24 h, 3 USDC = 30 days + Team Kit, 8 USDC = 90 days + Team Kit, by plain USDC transfer on Base; see payment_options), or pay per call over HTTP x402 ($0.02) at /v1/gha/pin.

    Annotations the server declares

    • readOnlyHint: true
    • openWorldHint: true
Only reads 1Show 1 read-only tool
  • list_rulesFREE.No findings

    Why it is in this group

    • The name "list_rules" says "list": a word for reading.
    Description, as the server wrote it
    FREE. The rule families CI Guard checks, with one line each, and the methodology.

    Annotations the server declares

    • readOnlyHint: true
    • openWorldHint: false

This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.

Exfil surface

Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.

No private-data, untrusted-input, outbound path in this server alone: no tool is tagged private-data.

Reads private data0
Reads files, mail, messages, databases, secrets, calendars or repositories.
Reads text others write5
Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
Can send data out3
Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.

Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.

Findings

Matches of fixed checks, with the text they matched.

3 warn1 info
  • The server marks "pin_actions" read-only (readOnlyHint: true), and its own name, description or input give the class "write".

    Matched text

    annotations: {"readOnlyHint":true,"openWorldHint":true}; The name "pin_actions" says "pin": a word for a change.

    What this check looks for
  • The server marks "payment_options" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

    Matched text

    annotations: {"readOnlyHint":true}; The name "payment_options" says "payment": a word that usually means money moves, which cannot be taken back.

    What this check looks for
  • The server marks "redeem_transfer_pass" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".

    Matched text

    annotations: {"readOnlyHint":true,"openWorldHint":true}; The name "redeem_transfer_pass" says "transfer": a word for a change that another call does not undo.

    What this check looks for
  • The description mentions "credentials".

    Matched text

    description: "...nd comments, artifact poisoning, leaked credentials, GITHUB_TOKEN permissions, secrets: inh..."

    What this check looks for

History

Changes to its tool definitions and to its remote endpoint, newest first.

No change seen yet. Its tools are rescanned every day, and a remote endpoint is looked at once a week.

Install

Pinned to version 1.1.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http ci-guard https://ravi-ci-guard.agent-arena-base.workers.dev/mcp
claude mcp add --transport http ci-guard https://ravi-ci-guard.agent-arena-base.workers.dev/mcp

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json
{
  "permissions": {
    "allow": [
      "mcp__ci-guard__list_rules"
    ],
    "ask": [
      "mcp__ci-guard__audit_sample",
      "mcp__ci-guard__peek_workflows",
      "mcp__ci-guard__audit_workflows",
      "mcp__ci-guard__pin_actions",
      "mcp__ci-guard__payment_options",
      "mcp__ci-guard__redeem_transfer_pass"
    ],
    "deny": []
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as ci-guard. Review it before you use it.

Capability tags behind 5 rules
  • mcp__ci-guard__audit_sample untrusted-input
  • mcp__ci-guard__peek_workflows untrusted-input, outbound
  • mcp__ci-guard__audit_workflows untrusted-input, outbound
  • mcp__ci-guard__pin_actions untrusted-input, outbound
  • mcp__ci-guard__redeem_transfer_pass untrusted-input

Pin

Keep the decision: fail CI when the tools change, follow every change, or show a badge.

Pin in CI

Fail a build when the tools you reviewed are not the tools the server serves today.

curl -s "https://radar.anlyon.com/api/v1/servers/dev.workers.agent-arena-base.ravi-ci-guard/ci-guard/verify?hash=89b23e605f2319bcc7c16feeab48f13587d50a780daa50ead0b7856e34ec3094" | jq -e '.changed == false'

Pinned hash 89b23e605f23, from Nov 21, 2026, 15:18 UTC.

Watch

Get told when its tools change: a new tool, a new description, a new class.

https://radar.anlyon.com/servers/dev.workers.agent-arena-base.ravi-ci-guard/ci-guard/changes.xml

An Atom feed. Works in any feed reader or Slack's RSS app.

Badge

Anlyon Radar badge for dev.workers.agent-arena-base.ravi-ci-guard/ci-guard
[![Anlyon Radar](https://radar.anlyon.com/badge/dev.workers.agent-arena-base.ravi-ci-guard/ci-guard.svg)](https://radar.anlyon.com/servers/dev.workers.agent-arena-base.ravi-ci-guard/ci-guard)

Source: MCP Registry. Synced Nov 21, 2026, 15:17 UTC.

Ravi CI Guard: GitHub Actions security audit MCP server: questions and answers

Answered from the scan above, not from the publisher's description.

Is the Ravi CI Guard: GitHub Actions security audit MCP server safe?

No scan can promise that. What its definitions show: 7 tools, 6 of which can change, send or delete something, and 4 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.

Can the Ravi CI Guard: GitHub Actions security audit MCP server delete data?

No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.

Can the Ravi CI Guard: GitHub Actions security audit MCP server do things that cannot be undone?

Yes. 2 tools can send, pay, publish or merge: payment_options and redeem_transfer_pass.

What tools does the Ravi CI Guard: GitHub Actions security audit MCP server have?

7 tools: 2 irreversible, 4 write, 1 read. payment_options, redeem_transfer_pass, audit_sample, peek_workflows, audit_workflows, pin_actions and list_rules.

Does the Ravi CI Guard: GitHub Actions security audit MCP server send data to outside services?

3 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): peek_workflows, audit_workflows and pin_actions.

Has the Ravi CI Guard: GitHub Actions security audit MCP server changed its tools?

Not since Anlyon Radar started tracking it. Last scanned Nov 21, 2026.