GongCAUTION
A framework for building agentic apps
Overview
A framework for building agentic apps
d31edd75948aOBSERVED · 2026-10-14Host compatibility
What the documentation claims. We have not run a compatibility test.
| Host | Status | Notes |
|---|---|---|
| cursor | mentioned |
What it tells the agent
The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.
--- name: gong description: > Search sales call recordings, transcripts, and participants via Gong. Use this skill when the user asks about sales calls, customer conversations, or call transcripts. --- # Gong Integration (Sales Calls) ## Connection - **Base URL**: `GONG_API_BASE` if configured, otherwise `https://api.gong.io/v2` - **Auth**: HTTP Basic — `Base64($GONG_ACCESS_KEY:$GONG_ACCESS_SECRET)` - **Env vars**: `GONG_ACCESS_KEY`, `GONG_ACCESS_SECRET`, optional `GONG_API_BASE` - **Caching**: 10-minute in-memory cache, max 120 entries ## Server Lib & Action - **File**: `server/lib/gong.ts` - **Action**: `gong-calls` - **Bundled deep-dive action**: `account-deep-dive` for HubSpot + Gong account or deal investigations. ### Exported Functions | Function | Description | | ----------------------------- | ----------------------------------------------------------- | | `getCalls(filters?)` | List calls (cursor-paginated) | | `getCall(callId)` | Get single call detail | | `getCallTranscript(callId)` | Get one call transcript | | `getCallTranscripts(callIds)` | Batch-fetch call transcripts | | `getUsers()` | List Gong users | | `searchCalls(query, days)` | List + filter by title and external party name/email/domain | | `getCallDetail(callId)` | Get parties plus Gong brief/key points/outline | | `getEnrichedTranscript(id)` | Transcript mapped to Gong parties and external speakers | ### UI API Routes | Route | Description | | --------------------- | ----------------- | | `GET /api/gong/calls` | List/search calls | | `GET /api/gong/users` | List users | Use `gong-calls` for agent-facing Gong work. Do not call `/api/gong/*` directly from the agent. ## Script Usage ```bash # Recent calls with a customer, including call-content evidence pnpm action gong-calls --company="Example Inc" --days=180 --includeTranscripts=true --transcriptLimit=5 # HubSpot + Gong account/deal deep dive pnpm action account-deep-dive --query="Example Inc" --days=180 --gongLimit=10 --transcriptLimit=5 # Get compact transcript text pnpm action gong-calls --transcript=<callId> # Get the raw transcript payload only for debugging/export pnpm action gong-calls --transcript=<callId> --rawTranscript=true # List Gong users pnpm action gong-calls --users ``` ## Key Patterns & Gotchas - **IMPORTANT API endpoints**: - `GET /v2/calls` — lists calls (with `fromDateTime`, `toDateTime`, `cursor` params) - `POST /v2/calls` — **uploading/creating** calls (NOT listing). Using this for listing returns 400 errors about missing fields. - `POST /v2/calls/extensive` — detailed call data with party info - `POST /v2/calls/transcript` — get transcripts - **Search pattern**: List calls via `GET /v2/calls?fromDateTime=...`, then filter client-side by company name, domain, person, or email against title and external parties. The search path batches `/v2/calls/extensive` for party matching when the lightweight list response is missing parties. No server-side company name search. - **Transcripts**: `gong-calls --transcript=<callId>` returns compact extracted text by default. Set `rawTranscript=true` only for debugging/export; do not pass raw transcript payloads into `save-analysis`. - Raw Gong transcript payloads have `speakerId` (numeric), `topic` (string or null), `sentences` array with `start`/`end` (ms) and `text`. Speaker IDs need cross-referencing with call parties. - For deal/customer deep dives, call `account-deep-dive` first when HubSpot context matters. For Gong-only follow-up, set `includeTranscripts=true`; call metadata alone is not enough for objections, risks, sentiment, or next-step claims. - Region/hostname is configurable with `GONG_API_BASE`; omit it for the global endpoint. ## Complete-Coverage Transcript Scan When the question asks whether any call in a cohort mentions something, do not answer from call metadata, titles, briefs, or sampled transcript excerpts. Use a raw transcript corpus path: 1. Discover or stage the exact call IDs in scope with bounded filters and explicit coverage counts. 2. Prefer `provider-corpus-job` with `mode: "batch-search"` against `POST /calls/transcript`. 3. Use `batch.itemBodyPath: "filter.callIds"`, `batch.responseItemsPath: "callTranscripts"`, `batch.batchSize: 20`, `search.textPaths: ["transcript"]`, and `search.idPaths: ["callId"]`. 4. Join the stored hits back to deals/accounts with `run-code` or staged-dataset queries, then report cohort size, call IDs discovered, transcript records scanned, hits, errors, and quota gaps. Use `provider-api-catalog --provider=gong` to retrieve the reusable corpus recipe. A metadata scan over `/calls` or `/calls/extensive` is useful for call discovery, but it is not evidence that transcript text lacks a phrase.
Trust audit
CAUTIONgrade B · trust 89/100 Install with care. The audit found things worth knowing before you trust its output.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | WARN |
| L1 | Static analysis of the code | NA |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (5)
.agents/skills/an
.agents/skills/content-product-development
.agents/skills/design-exploration
.agents/skills/turn-into-app
.agents/skills/visual-edit
Gates applied: no_behavioural_pass.
d31edd75948afull audit observations/trust-audit/skill/builderio__gong.json · Report an issue / request a re-scanAudit history
Every audit this skill has had.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-14 | d31edd75948a | CAUTION | B | 89 | first audit |
Questions
What does the Gong skill do?
A framework for building agentic apps
Is Gong safe to install?
With care. The audit graded it B (89/100) and found 5 things worth knowing before you trust this skill, listed below with the exact line each was found on.
What can Gong access on my machine?
The audit observed no filesystem, network or shell use at all in its source.
Which assistants does Gong work with?
Its documentation mentions cursor. That is what the text claims, not a compatibility test we ran.
How current is this page?
The grade is for one exact copy of the source (d31edd75948a), read on 2026-10-14. The repository is watched, and a new audit runs when it changes — this is the first audit.
Add this badge to your README
Shows this page's audit result and links back to it. It updates when the audit does.
[](https://modelsatlas.com/skills/builderio/gong/)
Not your listing, or something out of date? Claim or update your listing.