observe-whatsapp
by gokapsoobserve-whatsapp helps diagnose WhatsApp operations in Kapso: message delivery, webhook retries, API errors, and number health. Use the observe-whatsapp skill to move from a failure signal to the right check fast, with CLI-first steps and API fallback for production troubleshooting.
This skill scores 76/100, which is a solid but not top-tier listing candidate. Directory users should see real operational value for WhatsApp troubleshooting—especially delivery investigation, webhook debugging, API error triage, and health checks—but should also expect to rely on Kapso access and the included scripts rather than a fully polished one-command workflow.
- Strong triggerability: the frontmatter explicitly says to use it for production message failures, webhook delivery problems, and WhatsApp health checks.
- Good operational depth: the repo includes 15 scripts plus reference playbooks for health, message debugging, and triage workflows.
- Useful progressive disclosure: the SKILL.md provides preferred Kapso CLI paths and fallback Node scripts for the same diagnostics.
- Adoption depends on Kapso setup: the skill assumes the Kapso CLI is installed and authenticated, with env-var fallback only if that is unavailable.
- No install command in SKILL.md, so users may need to infer setup and execution details from the workflow text and scripts.
Overview of observe-whatsapp skill
The observe-whatsapp skill helps you diagnose WhatsApp operations in Kapso: message delivery, webhook retries, API errors, and number health. It is built for support engineers, operators, and integrators who need to move from “something failed” to a concrete cause fast, without reading every repo file first.
Use the observe-whatsapp skill when you need to verify whether a message was sent, delivered, or read; determine why a webhook did not arrive; or confirm whether a WhatsApp configuration is degraded, blocked, or healthy. The main value is not just diagnostics, but a clearer triage path: it tells you what to inspect first and what evidence matters most.
What this skill is best for
observe-whatsapp is strongest for production troubleshooting, especially when the user has a message ID, phone number, webhook failure, or a failing health check. It is less about generic WhatsApp guidance and more about operational state in Kapso.
What makes observe-whatsapp different
This skill is workflow-driven and evidence-based. It includes direct Kapso CLI paths plus fallback scripts, so you can use it whether you are in a logged-in CLI session or working from API credentials. That makes observe-whatsapp for Observability more practical than a prompt that only describes symptoms.
When it is a good fit
Choose observe-whatsapp if you need to inspect delivery timelines, webhook attempts, health signals, or API logs. If you only need a canned response or a one-off explanation, a normal prompt may be enough; if you need repeatable diagnosis, this skill is a better fit.
How to Use observe-whatsapp skill
Install context and prerequisites
The intended install path is:
npx skills add gokapso/agent-skills --skill observe-whatsapp
For best results, use the Kapso CLI first: authenticate with kapso login, then run kapso status to confirm project access and available WhatsApp numbers. If CLI access is unavailable, the skill also supports API-based fallback via KAPSO_API_BASE_URL and KAPSO_API_KEY.
Turn a rough issue into a usable prompt
The skill works best when your prompt includes the smallest actionable facts: a phone number, message ID, webhook event, timestamp window, or the exact error text. For example, instead of “WhatsApp is broken,” use: “Use observe-whatsapp to check why wamid.HBgMMTIzNDU2Nzg5 stopped at delivered and whether webhook retries failed after 10:00 UTC.”
Workflow that usually gives the fastest answer
Start with the user-facing symptom, then narrow to one of four paths: message delivery, webhook delivery, error triage, or health check. In practice, the skill expects you to confirm the config first, then inspect the specific artifact second. That sequence reduces false leads and avoids guessing about the wrong phone number or conversation.
Files to read first
For install-time understanding, read SKILL.md first, then references/health-reference.md, references/message-debugging-reference.md, and references/triage-reference.md. If you need examples, inspect assets/health-example.json, assets/message-debugging-example.json, and assets/triage-example.json. If you want to understand the fallback tooling, review scripts/messages.js, scripts/message-details.js, scripts/webhook-deliveries.js, and scripts/whatsapp-health.js.
observe-whatsapp skill FAQ
Do I need the Kapso CLI to use observe-whatsapp?
No, but it is the preferred path. The skill is designed to work best with kapso status and the WhatsApp subcommands. If you cannot use the CLI, the scripts and API environment variables provide a fallback.
What input does observe-whatsapp need?
Give it identifiers, not just symptoms: message ID, display number or E.164 phone number, phone-number ID, timestamp range, webhook event name, or the raw error message. The more specific your input, the less the skill has to infer.
Is observe-whatsapp beginner friendly?
Yes, if you already know the basic shape of the issue. It is easier to use than a blank prompt because it tells you what to inspect first, but it still assumes you can provide at least one concrete WhatsApp artifact.
When should I not use this skill?
Do not use observe-whatsapp for non-WhatsApp observability tasks or for product-level strategy questions. It is also a poor fit when you have no IDs, no timestamps, and no access to logs or status signals.
How to Improve observe-whatsapp skill
Provide the exact failure point
The fastest way to improve output quality is to name where the workflow broke: “sent but not delivered,” “webhook received with Connection refused,” or “health is degraded and webhook verification failed.” That helps observe-whatsapp skip broad explanation and focus on the failing check.
Include the evidence the skill can verify
Strong inputs include a message ID, the phone number used, the time window, and any response payloads or error codes. For example, “Use observe-whatsapp to inspect webhook deliveries for whatsapp.message.received between 10:00 and 10:15 UTC; last attempt returned 502” is far better than “webhooks are down.”
Use the reference files to sharpen first-run output
If the first answer is too generic, compare it with the interpretation rules in references/health-reference.md and the troubleshooting logic in references/triage-reference.md. These references show what should be treated as critical, degraded, or retryable, which improves both the diagnosis and the wording you return to users.
Iterate with one change at a time
If the initial result is incomplete, refine the prompt with one new fact: a different ID, a narrower timestamp window, or the exact failing check. For observe-whatsapp, that usually works better than restating the whole problem because the skill is built around tracing a single operational path to the root cause.