Imposter Smasher for Events and Meetings

Prepare a user for an upcoming meeting or event by listing next-day calendar events, prompting for a selection, researching the topic and participants, produ...

Security Scan

The skill's files and runtime instructions are consistent with a meeting-prep/orchestration tool that delegates calendar lookup, web research, and audio rendering to external connectors; nothing in the package indicates unexplained or disproportionate access requests or hidden exfiltration.

Purpose & Capability

Name/description match the manifest and scripts. The skill is an orchestrator that lists calendar events, collects research, synthesizes an executive summary, and produces an audio briefing; the included helper scripts are narrowly scoped to formatting, runtime estimation, and manifest generation and align with that purpose.

Instruction Scope

SKILL.md confines actions to: fetch next-day calendar events, call Contextual.ai for web research, and call ElevenLabs/Chatterbox for audio. It has clear failure modes (ask user for pasted event details) and guardrails (no outreach, no impersonation). The instructions do not ask the agent to read arbitrary host files, search system paths, or exfiltrate unrelated environment variables.

Install Mechanism

No install spec provided (instruction-first), so nothing is downloaded or executed beyond the included, small helper scripts. The scripts are plain Python/Bash with no remote downloads or obfuscation. This is a low-risk install footprint.

Credentials

SKILL.md depends on external services (calendar, Contextual.ai, ElevenLabs/Chatterbox) but declares no required env vars. This is coherent if the hosting agent/platform supplies those connectors/credentials; however, users should confirm which connector credentials will be used and where they are stored because the skill will route meeting details and research requests to external services.

Persistence & Privilege

No 'always: true' privilege requested and default autonomous invocation is unchanged. The skill writes only its own output artifacts (executive_summary.md, audio_script.txt, audio_manifest.json) to an outdir and does not modify other skills or system-wide configs.

Assessment

This skill appears internally consistent, but before installing verify the following: (1) which calendar and research/audio connectors the agent will use and that you trust their handling of meeting content; (2) whether API keys or account access will be provided to Contextual.ai or ElevenLabs/Chatterbox (sensitive meeting details could be sent to those services); (3) storage and retention of generated audio files and any transcripts; and (4) test the skill with a non‑sensitive meeting first. If you need stricter privacy, redact sensitive details before providing events or avoid using external research/audio services.

Imposter Smasher

Purpose

Imposter Smasher is an orchestrator skill for high-stakes meeting prep. It compiles next-day meetings, lets the user pick one, researches the topic and participants with credible sources, then delivers:

1. A concise executive summary.
2. A fully produced 3-5 minute audio briefing.

Use This Skill When

• The user wants prep for a meeting happening tomorrow.
• The user wants a confidence-boosting brief before a customer, investor, executive, or partner call.
• The user wants both text and audio output.

Hard Dependencies

• Calendar access (to list and inspect next-day events).
• Web research via Contextual.ai.
• Audio generation via ElevenLabs or Chatterbox.

Out Of Scope

• Sending emails/messages on behalf of the user.
• Booking, modifying, or cancelling meetings.
• Unsupported speculation or rumor-based profiling.
• Long dossiers beyond prep needs.
• Live in-meeting copilot behavior.
• Research that cannot be grounded in credible sources.

Inputs To Collect

• Timezone (if unclear).
• Target date (default: next day in user timezone).
• Preferred audio engine (ElevenLabs or Chatterbox).
• Optional persona/tone for briefing voice.

Orchestration Workflow

1. Fetch next-day calendar events.
2. Present a numbered shortlist with title, start time, organizer, and attendees.
3. Ask user to choose one event.
4. Extract research targets:
   • Meeting topic and company/domain context.
   • Participants and their roles.
   • Strategic risks, opportunities, and likely questions.
5. Delegate specialized subtasks where possible:
   • Calendar retrieval/parsing to calendar-capable tooling.
   • Web research and source collection to Contextual.ai.
   • Audio rendering to ElevenLabs/Chatterbox integration.
6. Synthesize an executive summary using references/executive-summary-template.md.
7. Build final audio script with references/audio-brief-template.md.
8. Generate a produced 3-5 minute audio file and return artifact paths/links.
9. Return concise prep package:
   • Executive summary.
   • Top participant notes.
   • Risks/questions checklist.
   • Audio file location and duration.

Quality Bar

• Cite only credible, attributable sources.
• Distinguish facts vs inferences.
• Keep briefing actionable and concise.
• Target spoken runtime between 180 and 300 seconds.
• If evidence is weak, explicitly say so and reduce confidence.

Failure Handling

• If no calendar access: ask for pasted event details and continue.
• If research fails: provide a minimal brief with explicit gaps and retry options.
• If audio generation fails: provide final narration script and engine-specific retry command.

Detailed References

• Workflow rubric: references/workflow-rubric.md
• Source credibility rules: references/source-credibility-rubric.md
• Executive summary template: references/executive-summary-template.md
• Audio script template: references/audio-brief-template.md
• Concise implementation notes: references/implementation-notes.md

Helper Scripts

• scripts/build_briefing_packet.py: compile event + research notes into summary and narration draft.
• scripts/estimate_runtime.py: estimate spoken duration and validate 3-5 minute target.
• scripts/validate_skill.sh: basic scaffold validation.