Skills/Glama/muster

muster@r1

CAUTION· install only

Publisher claims: "Enables Claude Code sessions to discover and exchange live messages on one machine or across paired machines, waking idle sessions via hooks without dangerous flags, with durable message storage and secure TLS pairing." Observed in sandbox: calls github.com only, touches nothing outside the project, writes no memory.

Registry GlamaPublisher castor-o7 · publishing for 3 daysInstalls 0Last detonated 3d ago
What would fix thisWatchDispute verdict

Coverage

Install scripts ran; no agent task scenario for this bundle.
Tested
install scripts, with decoy credentials in place · second launch with the clock 45 days ahead · every file read, file written and host contacted by what ran
Not tested
use of the skill by an agent

What this means for the label: Caution here covers what ran. The server's tools were not exercised with a real credential, so nothing about their behavior is asserted; the credential summary below says what to do with that.

Before you give it your token

It will ask for
D_TOKENE_TOKENL_TOKENLATE_TOKENLINK_TOKENLOCAL_TOKENMTOKENREVOKED_TOKEN or --max-tokensConfiguration, not credentials: BASE_PATH, CLAUDE_CONFIG_DIR, CODEX_HOME, FAKE_ID, MUSTER_ACK_LEASE_SECS, MUSTER_BASE_URL, MUSTER_BIND, MUSTER_CHROMIUM_PATH, MUSTER_EVAL_REPRESENTATION, MUSTER_EVAL_SCENARIO, MUSTER_HARNESS, MUSTER_HOST, MUSTER_IDENTITY, MUSTER_INTERNAL_BIND, MUSTER_OPERATOR, MUSTER_PLAYWRIGHT_MODULE, MUSTER_TEST_SHIM, MUSTER_UI_SCREENSHOT_DIR, MUSTER_WAITER_LOCK_WAIT_SECS, OTHER_ID, PROGRAMDATA, RUST_LOG, USERNAME
Where it talks to
muster.example (its own)Also apple.com, rustup.rs, evil.example — not the publisher's domain; check what for before connecting.
What the honeypot saw
no credential files read · reached only github.com
What we could not test
use of the skill by an agent

Connect it with a scoped credential, under the guard, and read the third-party hosts first. The code names apple.com, rustup.rs, evil.example as well as its own vendor. The guard injects D_TOKEN, E_TOKEN, L_TOKEN, LATE_TOKEN, LINK_TOKEN, LOCAL_TOKEN, MTOKEN, REVOKED_TOKEN only into requests to muster.example (and its subdomains) and refuses the rest.

$vet-skills run --secrets D_TOKEN=muster.example,E_TOKEN=muster.example,L_TOKEN=muster.example,LATE_TOKEN=muster.example,LINK_TOKEN=muster.example,LOCAL_TOKEN=muster.example,MTOKEN=muster.example,REVOKED_TOKEN=muster.example -- <the server's command>

Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.

Is muster@r1 safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios; install scripts ran; no agent task scenario for this bundle) on 2026-11-11 and scored it 73/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: Destructive command against a filesystem root or device; Destructive file system command detected. Summary: calls github.com only, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 1 s · 2 scenarios · 0 deviations
t+CategoryObserved action
0.3snetworkCONNECT tcp://github.com:443
0.3sprocessexec /usr/bin/uname
0.3sprocessexec /usr/bin/curl
0.3sprocessexec /usr/bin/which
0.3sprocessexec /usr/bin/grep
0.3sprocessscript backend runs setup only
0.3sprocessscript backend runs setup only
0.4sprocesssecond launch, clock +45 days, CI unset
0.6snetworkCONNECT tcp://github.com:443
0.6sprocessexec /usr/bin/uname
0.6sprocessexec /usr/bin/curl
0.6sprocessexec /usr/bin/grep
0.6sprocessexec /usr/bin/which

Static findings

MEDIUM
Python subprocess execution detectedscripts/idle-turns.py:146 · Python subprocess execution detecteddone = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedscripts/sandbox/run.py:277 · Python subprocess execution detected'codex_version': subprocess.check_output([self.args.codex, '--version'], text=True).strip(),Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedscripts/test-gate-e2e.sh:307 · Python subprocess execution detectedp = subprocess.Popen([shim], stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.DEVNULL, env=env, text=True, bufsize=1)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill commandscripts/test-gate-e2e.sh:26 · Process kill command — agent can terminate processestrap 'kill $DPID 2>/dev/null; wait $DPID 2>/dev/null' EXITFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedscripts/sandbox/test_run.py:61 · Python subprocess execution detectedsubprocess.run([os.environ['MUSTER_TEST_SHIM'], 'codex-setup', '--no-install', '--codex-home', str(home), '--daemon', 'ws://127.0.0.1:45678'], check=True, capture_output=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
File permission modification commandscripts/install-codex.sh:121 · File permission modification command — agent changes file ownership/permissionschmod +x "$bin"Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
File type combined with upload actionscripts/sandbox/campaign.py:56 · File type combined with upload action — agent may exfiltrate local data filessummary = json.loads((output / 'summary.json').read_text()) if (output / 'summary.json').exists() else {}Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
MEDIUM
File type combined with upload actioncrates/shim/tests/fixtures/codex-0.163.0/README.md:13 · File type combined with upload action — agent may exfiltrate local data files- `user-prompt-submit.json`, `post-tool-use.json`: hook input.Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
CRITICAL
Destructive command against a filesystem root or devicescripts/test-setup.sh:184 · Destructive command against a filesystem root or devicerun --tag "v1;rm -rf /"Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
HIGH
Destructive file system command detectedscripts/test-setup.sh:184 · Destructive file system command detectedrun --tag "v1;rm -rf /"Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedscripts/test-gate-e2e.sh:391 · Python subprocess execution detectedp = subprocess.Popen([shim], stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=open(env_home + "/shim.log", "a"), env=env, text=True, bufsize=1)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
File permission modification commandscripts/setup.sh:232 · File permission modification command — agent changes file ownership/permissionschmod +x "$tmp/$b"Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
Process kill commandscripts/test-restart-e2e.sh:106 · Process kill command — agent can terminate processeskill -TERM $DPID; wait $DPID 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/test-links-e2e.sh:22 · Process kill command — agent can terminate processestrap 'kill $LPID $DPID 2>/dev/null; wait 2>/dev/null' EXITFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/test-gate-e2e.sh:235 · Process kill command — agent can terminate processessleep 1; kill -0 $LATE_PID 2>/dev/null; ALIVE=$?Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/test-links-e2e.sh:226 · Process kill command — agent can terminate processeskill $DPID; wait $DPID 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
File permission modification commandscripts/test-sign-release.sh:38 · File permission modification command — agent changes file ownership/permissionschmod +x "$WORK/fake/gh"Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
File permission modification commandscripts/test-setup-e2e.sh:46 · File permission modification command — agent changes file ownership/permissionschmod +x "$WORK/fake"/*Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
File permission modification commandscripts/test-setup.sh:54 · File permission modification command — agent changes file ownership/permissionscase " $* " in *" $b "*) "$MAKE_FAKE" "$b" > "$root/bin/$b"; chmod +x "$root/bin/$b" ;; esacFix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
File permission modification commandscripts/test-setup.sh:70 · File permission modification command — agent changes file ownership/permissionschmod +x "$FAKE"/* "$WORK/make-fake"Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
File type combined with upload actionscripts/test-restart-e2e.sh:36 · File type combined with upload action — agent may exfiltrate local data filespg_ctl -D $WORK/pg -l $WORK/pg.log -o "-p $PGPORT -k $SOCK -c listen_addresses=''" -w start >/dev/null || { echo "postgres didn't start"; cat $WORK/pg.log; exit 1; }Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
MEDIUM
Process kill commandscripts/test-restart-e2e.sh:37 · Process kill command — agent can terminate processestrap 'kill $DPID 2>/dev/null; wait 2>/dev/null; pg_ctl -D $WORK/pg -m immediate stop >/dev/null 2>&1; rm -rf $SOCK' EXITFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/test-links-e2e.sh:110 · Process kill command — agent can terminate processeskill $LPID; wait $LPID 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/test-gate-e2e.sh:238 · Process kill command — agent can terminate processeskill $LATE_PID 2>/dev/null; wait $LATE_PID 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/test-links-e2e.sh:263 · Process kill command — agent can terminate processeskill $DPID; wait $DPID 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
File permission modification commandscripts/test-rollout.sh:69 · File permission modification command — agent changes file ownership/permissionsexit 1' >"$WORK/bad"; chmod 755 "$WORK/bad"Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
Process kill commandscripts/test-gate-e2e.sh:237 · Process kill command — agent can terminate processescheck "unpair closes an open long-poll within seconds" '[[ $ALIVE == 0 ]] && ! kill -0 $LATE_PID 2>/dev/null'Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedscripts/test-restart-e2e.sh:56 · Python subprocess execution detectedp = subprocess.Popen([shim], stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.DEVNULL, env=env, text=True, bufsize=1)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill commandscripts/test-restart-e2e.sh:29 · Process kill command — agent can terminate processestrap 'kill $DPID 2>/dev/null; wait 2>/dev/null; drop_run_schema' EXITFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/test-gate-e2e.sh:196 · Process kill command — agent can terminate processeskill $HELD 2>/dev/null; wait $HELD 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/test-links-e2e.sh:197 · Process kill command — agent can terminate processeskill $DPID; wait $DPID 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedscripts/test-links-e2e.sh:57 · Python subprocess execution detectedp = subprocess.Popen([shim], stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.DEVNULL, env=env, text=True, bufsize=1)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
File permission modification commandscripts/test-setup.sh:202 · File permission modification command — agent changes file ownership/permissions"$MAKE_FAKE" muster-shim >"$WORK/stale/muster-shim"; chmod +x "$WORK/stale/muster-shim"Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
File type combined with upload actionscripts/test-gate-e2e.sh:530 · File type combined with upload action — agent may exfiltrate local data files…0.1 16456 2>/dev/null && break; sleep 0.2; done curl -s -m 20 -D $WORK/stop-head.txt -o $WORK/stop-body.txt -H "Authorization: Bearer $LOCAL_TOKEN" -X POST "http://127.0.0.1:16456/hook/wait?timeout=15" -d '{"session_id":"a0000000-0000-…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07

Cost to the agent

Instructions loaded
≈32 tok
1 file · 81 in bundle

Tokens estimated at four characters each. Skills load their instruction text when triggered; the figure is what that costs.

Known vulnerabilities

Vulnerability advisories (OSV) cover npm and PyPI packages; this registry has no advisory database.

What would make this Verified

7 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%
  1. Destructive command against a filesystem root or device≈ +9 on the totalBuild commands from fixed arguments; do not interpolate untrusted strings into a shell.
  2. Unclaimed publisher≈ +8 on the totalClaim castor-o7: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
  3. Destructive file system command detected≈ +4 on the totalBuild commands from fixed arguments; do not interpolate untrusted strings into a shell.
  4. Python subprocess execution detected≈ +2 on the totalBuild commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
  5. Process kill command≈ +2 on the totalBuild commands from fixed arguments; do not interpolate untrusted strings into a shell.
  6. File permission modification command≈ +2 on the totalRequest the narrowest permission that works; replace wildcards with the exact paths or hosts.
Publish a new version to Glama, then queue it here. Not your skill? Claim the handle to get verdict changes by email.Claim castor-o7Propose a patch to the maintainer