Skills/npm/summarizing-meetings

summarizing-meetings@1.0.0

VERIFIED· install only

Publisher claims: "Agent skill: turn a raw meeting transcript (Granola, Zoom, Meet, any ASR) into a clean record and a follow-up with action items." Observed in sandbox: no network, touches nothing outside the project, writes no memory.

Registry npm (MCP server)Publisher vstlmkh · publishing for 30 daysInstalls 220Last detonated 38 min ago
$claude mcp add summarizing-meetings -- npx -y @vstlmkh/summarizing-meetings
WatchReport an issue

Coverage

Bundled scripts ran with no arguments; no agent task scenario for this bundle.
Tested
1 of 1 bundled script run with no arguments, with decoy credentials in the environment · every file read, file written and host contacted by what ran
Not tested
use of the skill by an agent

Is [email protected] safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios; bundled scripts ran with no arguments; no agent task scenario for this bundle) on 2026-11-20 and scored it 90/100 (Verified). In what ran, nothing we could detect went wrong: no undeclared network egress, no credential reads, no persistent instruction writes. What it does with a real credential was not observed; the static findings and the publisher carry the rest of the score. Summary: no network, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 0 s · 2 scenarios · 0 deviations
t+CategoryObserved action
0.0sprocessno install scripts in bundle
0.0sprocessbundled script run in the sandbox with no arguments
0.1sprocess1 of 1 bundled script run with no arguments
0.1sprocessscript backend runs setup only
0.1sprocessscript backend runs setup only
0.1sprocessno install scripts to re-run

Static findings

MEDIUM
npx invocation of non-well-known packagebin/install.js:3 · npx invocation of non-well-known package — may execute untrusted code// Usage: npx @vstlmkh/summarizing-meetings [--dir <skills-dir>] [--force]Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
Process kill commandSKILL.md:216 · Process kill command — agent can terminate processesdeadline. Soft commitments inflate the list and kill trust in it.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
npx invocation of non-well-known packagebin/install.js:28 · npx invocation of non-well-known package — may execute untrusted codeconsole.log(`Usage: npx @vstlmkh/${SKILL_NAME} [--dir <skills-dir>] [--force]Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
LOW
Prompt injectionevals/files/sample-granola-transcript.txt:7 · Prompt injection — known jailbreak technique reference · informational: not an instruction fileMe: ok. let me take a look. actually no better let Dan build the prototype on the workerFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST01
LOW
Prompt injectionreferences/transcript-cleanup.md:146 · Prompt injection — known jailbreak technique reference · informational: not an instruction fileMe: ok. let me take a look. actually no better let Dan build a prototypeFix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionreferences/followup-formats.md:70 · Prompt injection — known jailbreak technique reference · informational: not an instruction file**Actions.** @Dan — prototype — by 24.10Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionreferences/followup-formats.md:104 · Prompt injection — known jailbreak technique reference · informational: not an instruction filesomeone absent is not acceptance — flag it: `@Dan (not on the call)`.Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionevals/output-conformance.json:19 · Prompt injection — known jailbreak technique reference · informational: not an instruction file"The prototype action item is owned by Dan, not Mark, despite Mark first saying 'let me take a look'",Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST01
LOW
Prompt injectionreferences/transcript-cleanup.md:165 · Prompt injection — known jailbreak technique reference · informational: not an instruction fileout of the decisions list), the reassignment from Mark to Dan (the owner changedFix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionreferences/followup-formats.md:34 · Prompt injection — known jailbreak technique reference · informational: not an instruction file**Participants:** Anna, Mark, DanFix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionevals/output-conformance.json:5 · Prompt injection — known jailbreak technique reference · informational: not an instruction file…es, migration of legacy orders ESM-15410. Participants: Mark (Me), Anna (Them), Dan is mentioned but was not on the call. Meeting date 17.10.2026. Cleaned transcri…Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST01
LOW
Prompt injectionevals/output-conformance.json:20 · Prompt injection — known jailbreak technique reference · informational: not an instruction file"Dan is marked as not having been on the call, since he never accepted the task",Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST01
LOW
Prompt injectionreferences/followup-formats.md:44 · Prompt injection — known jailbreak technique reference · informational: not an instruction file- [ ] @Dan — build a retry prototype on the worker — by 24.10Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionreferences/transcript-cleanup.md:159 · Prompt injection — known jailbreak technique reference · informational: not an instruction file**Mark.** Let's have Dan build a prototype instead.Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01
LOW
Prompt injectionevals/output-conformance.json:21 · Prompt injection — known jailbreak technique reference · informational: not an instruction file"The Anna action item carries the explicit date 20.10, while the Dan item marks 'by next week' as an unfixed deadline",Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST01
LOW
Prompt injectionreferences/followup-formats.md:88 · Prompt injection — known jailbreak technique reference · informational: not an instruction file| 1 | Retry prototype on the worker | Dan | 24.10 |Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST01

Cost to the agent

Instructions loaded
≈2,935 tok
1 file · 11 in bundle

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

Known vulnerabilities

No advisories in OSV for @vstlmkh/summarizing-meetings@1.0.0 as of 3 min ago. This is the package itself, not its dependencies.