Skip to content
Skills Library
Programming · Analysis

sempera-debug-call

/sempera-debug-call

Work out why a Sempera call or conversation went wrong, from the call record, transcript, tool timeline and webhook delivery. Use when a call failed, an agent said something wrong, or a webhook did not arrive.

Repository
sempera-ai ↗
Licence
MIT
Last repository change
05.11.2026

Assessment and security

90/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content25 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 05.11.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Work out why a Sempera call or conversation went wrong, from the call record, transcript, tool timeline and webhook delivery. Use when a call failed, an agent said something wrong, or a webhook did not arrive.

Original author description

Work out why a Sempera call or conversation went wrong, from the call record, transcript, tool timeline and webhook delivery. Use when a call failed, an agent said something wrong, or a webhook did not arrive.

How it works

  1. Steps
  2. Rules
Original SKILL.md headings · 24 lines, 385 words
SKILL.md preview
---
name: sempera-debug-call
description: Work out why a Sempera call or conversation went wrong, from the call record, transcript, tool timeline and webhook delivery. Use when a call failed, an agent said something wrong, or a webhook did not arrive.
---

# Debug a call

Read first, change nothing until you know the cause.

## Steps

1. **Get the record.** `GET /v1/calls/{call_id}` (MCP `get_call`; CLI `sempera calls get <id>`). Note the status, direction, duration and the agent version that took it. If you only have an agent, `GET /v1/calls?agent_id=...` lists newest first.
2. **Read the conversation.** `GET /v1/sessions/{call_id}` (`get_conversation`) gives the turns, the tool timeline, captured variables, outcome and the version that spoke. Find the first turn where the call went wrong.
3. **Classify.**
   - *Wrong words*: the instructions or knowledge of that version are at fault. Read that version with `GET /v1/agents/{agent_id}/versions/{version_id}`, reproduce with a typed turn, `POST /v1/agents/{agent_id}/preview` with `{"message": "..."}` (its `version` field defaults to `draft`), fix the draft, and follow `sempera-build-agent`.
   - *A tool failed*: the tool timeline shows the failing step. Fix the tool or the system it calls; do not change the agent's words to hide it.
   - *No call happened*: for an outbound call, read the refusal. Common ones are outside calling hours (`409`, `outside_sending_hours`), the number on the do-not-contact list, a reached outbound cap (`429 rate_limited` with `retry_after_s`) and a missing scope (`403 insufficient_scope`). These are working as designed; report them.
   - *Call not stored*: the number may not be bound to the agent. Check `GET /v1/phone-numbers?agent_id=...`.
4. **A webhook did not arrive.** `GET /v1/webhooks/{id}/deliveries` is the delivery log, one row per attempt. Check the endpoint is enabled and subscribed to the event, and that your server answered `2xx` within 10 seconds. `POST /v1/webhooks/{id}/test` sends one signed `webhook.test`.
5. **A signature does not verify.** The header `X-Sempera-Signature` is `t=<unix seconds>,v1=<hex>`, the HMAC-SHA256 of `"<t>." + raw_body` keyed with the endpoint secret. Verify the exact raw bytes before parsing JSON. To check a captured delivery locally: `sempera webhooks verify --signature '<header>' --body-file body.json` with `SEMPERA_WEBHOOK_SECRET` set, or the MCP tool `verify_webhook_signature`. Delivery is at least once: drop repeats by `X-Sempera-Event-Id`.
6. **Ask the user for the missing piece, not for a larger key.** If a read is refused for scope, say which scope.

## Rules

- Quote the API's error `code` and `request_id`; they are what support needs.
- Transcripts hold personal data: summarise, do not paste them elsewhere.
- Do not place a new call to reproduce a problem unless the user names the person and confirms (`sempera-send-call`).

Versions

The database has no other skills named sempera-debug-call.

Database snapshot: 13.11.2026 · package version d68acf0ceeb8 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100sempera-build-agent
    Create a Sempera AI phone agent, edit its draft (instructions, voice, language), try it with a typed preview and publish it. Use when the user wants to build, change or publish a Sempera agent.
  • 90sempera-read-conversations
    List and read Sempera calls and conversations with their transcripts. Use when the user asks what callers said, how a call went, or for recent activity on an agent.
  • 100debug-lab
    Find the real cause of hard bugs, slowdowns, flaky failures, broken live workflows and stuck AI-agent runs, then fix at the source and prove it end to end. Use for "debug this", "diagnose", "why is this failing", "it's broken", "throws", "error", "stack trace", "test failing",…
  • 90shopify-webhook-topic
    Add or change handling of a Shopify webhook topic in Order Hub (orders/*, customers/*, refunds/*, products/*, compliance topics). Use when the user asks to "handle a new webhook", "subscribe to X", "process refunds/create", or changes how a webhook payload is stored.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.