avoma-skill
v1.0.0
Published
Avoma API skill — reusable bash scripts for meeting intelligence, transcripts, scorecards, engagement analytics, and revenue intel. Installable as global CLI commands or as a Claude Code / OpenClaw agent skill.
Maintainers
Readme
Avoma Skill
Reusable bash scripts for the Avoma meeting intelligence API. Built for AI agents (OpenClaw, Claude Code, Codex) but works from any terminal.
What it does: Pull meeting transcripts, AI notes, call recordings, scorecard evaluations, engagement analytics, and revenue intelligence from Avoma — without rewriting curl commands or burning tokens figuring out the API every time.
20 scripts covering every Avoma v1 endpoint. Built by Claudio and Vito at RBP Consulting.
Install
As a Claude Code / OpenClaw skill:
npx avoma-skillCopies SKILL.md, scripts/, and references/ into .claude/skills/avoma-skill/ in the current directory.
As global CLI commands (adds avoma-recent, avoma-meetings, avoma-transcript, etc. to your PATH):
npm install -g avoma-skill
avoma-recent 7Or just clone the repo and run the scripts directly with bash scripts/avoma-*.sh — see below.
Quick Start
1. Get your Avoma API key
Follow Avoma's API Integration guide to generate your Client Key.
2. Set it in your environment
# Add to your .env or export directly
export AVOMA_API_KEY="your-client-key-here"The scripts also check ~/.openclaw/.env, ~/.openclaw/workspace/.env, and openclaw.json automatically.
3. Run a script
# Quick lookup: what meetings happened this week?
bash scripts/avoma-recent.sh 7
# Pull everything for a meeting at once — transcript, notes, insights, snippets
bash scripts/avoma-meeting-full.sh <MEETING_UUID> ./output/client-name/
cat ./output/client-name/transcript.txt
# Get AI notes in markdown
bash scripts/avoma-notes.sh <MEETING_UUID> --format markdown
# Engagement analytics for Q1
bash scripts/avoma-engagement.sh 2026-01-01 2026-03-31 --summaryScripts
Workflow Helpers (Start Here)
| Script | What it does |
|--------|-------------|
| avoma-recent.sh | Quick list of recent meetings with UUIDs — avoma-recent.sh 7 for the last week |
| avoma-meeting-full.sh | One-shot pull: transcript + notes + insights + snippets + sentiments → saved to disk |
Core Meeting Operations
| Script | What it does |
|--------|-------------|
| avoma-meetings.sh | List meetings by date range with filters (calls-only, external-only, by attendee, by CRM account) |
| avoma-meeting.sh | Get a single meeting by UUID |
| avoma-transcript.sh | Get transcript for a meeting, by UUID, or list by date range |
| avoma-notes.sh | Get AI-generated notes (markdown, json, or html) |
| avoma-insights.sh | Get meeting insights (AI notes + keywords + speaker stats) |
| avoma-snippets.sh | Get meeting highlights — AI-generated or user-created |
| avoma-recordings.sh | Get signed audio/video download URLs (valid 5 days) |
| avoma-sentiments.sh | Get emotional tone analysis over the meeting timeline |
| avoma-segments.sh | Get meeting segments (intro, demo, pricing, next steps, etc.) |
Calls
| Script | What it does |
|--------|-------------|
| avoma-calls.sh | List calls by date range (inbound/outbound filter), or get by external ID |
Scorecards & Engagement
| Script | What it does |
|--------|-------------|
| avoma-scorecards.sh | List scorecard templates |
| avoma-scorecard-evals.sh | List scorecard evaluations (filter by date, meeting, user, scorecard) |
| avoma-engagement.sh | Engagement analytics — who's listening, sharing, commenting on meetings |
Revenue Intelligence (Beta)
| Script | What it does |
|--------|-------------|
| avoma-revenue-intel.sh | CRM-linked timeline of meetings/calls/emails per deal or account |
Configuration & Metadata
| Script | What it does |
|--------|-------------|
| avoma-users.sh | List org users |
| avoma-meeting-types.sh | List/create/delete meeting types (purposes) |
| avoma-meeting-outcomes.sh | List/create/delete meeting outcomes |
| avoma-smart-categories.sh | List smart categories (keyword/prompt tracking) |
| avoma-templates.sh | List note templates |
How It Works
All scripts source scripts/avoma-config.sh which handles:
- Auth: Reads
AVOMA_API_KEYfrom environment,.envfiles, oropenclaw.json - Rate limits: 60 requests/minute — errors print clearly on 429
- Error handling: HTTP status detection with clear error messages
- Date formatting: Accepts
YYYY-MM-DD(auto-converts to RFC3339 UTC)
Common Patterns
SCRIPTS=./scripts
# Quick lookup: recent meetings → grab a UUID
bash $SCRIPTS/avoma-recent.sh 7
# Pull everything for a meeting at once (no repeated API calls)
bash $SCRIPTS/avoma-meeting-full.sh <UUID> ./output/client-name/
cat ./output/client-name/transcript.txt
# All external meetings this month with transcripts ready
bash $SCRIPTS/avoma-meetings.sh 2026-04-01 2026-04-30 --external-only --page-size 100 \
| jq '.results[] | select(.transcript_ready==true) | {uuid, subject, start_at}'
# Scorecard evaluations for a specific rep
bash $SCRIPTS/avoma-scorecard-evals.sh --user [email protected] --from 2026-04-01 --to 2026-04-30
# Revenue engagement timeline for a deal
bash $SCRIPTS/avoma-revenue-intel.sh opportunity 2026-01-01 --entity-id 56272392021 --interval week
# All snippets (highlights) from a meeting
bash $SCRIPTS/avoma-snippets.sh <MEETING_UUID> --ai-onlyFor OpenClaw Users
Drop the skills/avoma/ directory into your workspace skills/ folder. OpenClaw will pick up the SKILL.md automatically and reference it when you ask about Avoma.
API Reference
Full OpenAPI v1 spec is in references/openapi-v1.yml for when you need to check field names, query params, or response shapes.
- Rate limit: 60 requests per minute. Scripts will tell you when you hit it.
- Dates: All UTC, RFC3339 format. Scripts accept
YYYY-MM-DDfor convenience. - Pagination: Most list endpoints return
{count, next, previous, results}. Use--page-size 100for max page size.
Contributing
All changes go through PRs. Only the repo owner merges.
If you add a script, follow the pattern:
- Source
avoma-config.shat the top - Use
avoma_curl METHOD /endpoint/for all API calls - Add a usage comment block with examples
- Add it to the table in SKILL.md and this README
- Test against the live API before opening the PR
See CLAUDE.md for AI agent instructions and AGENTS.md for the full operating guide.
API Key Management
Each user/agent needs their own Avoma API key. Keys are never committed to this repo.
Where to store it:
- OpenClaw agents:
AVOMA_API_KEY=xxxin~/.openclaw/.env - Claude Code / terminal:
export AVOMA_API_KEY=xxx - The scripts check three locations automatically (env var →
.envfiles →openclaw.json)
License
MIT. Get your own API key from Avoma — don't share keys.
