Works with the AI Agents you already use
Timed Out? Check Before You Resend
Use when a send, upload or submit timed out and you can't tell if it went through. Checks the evidence against the original request, then gives a status and one safe next step, without sending twice.
Free
Timed Out? Check Before You Resend
Example session with this skill installed
Reconcile the status for Request REQ-992. Target: partner-api-01. Sent at 10:05 UTC. Log shows a 502 Bad Gateway at 10:05:02, but the DB shows a partial record created at 10:05:01.
- Read your context and instructions
- Compiled the timed out check
I have analyzed the provided logs and database records for Request REQ-992. While the transport layer reported a failure, the database state indicates the request was processed by the target system.
Delivery Reconciliation Record: REQ-992
| Field | Details |
|---|---|
| Original Request | ID: REQ-992 | Target: partner-api-01 | Submit Time: 10:05:00 UTC |
| Fingerprint | Not provided (using ID matching only) |
| Last Observation | 10:05:02 UTC | Entry Point: API Gateway | Status: 502 Bad Gateway |
| Conclusion | Accepted |
| Basis | Database record creation at 10:05:01 UTC confirms the application layer received and persisted the intent before the gateway timed out. |
| Gaps | Lack of byte fingerprint prevents verification that the DB record content exactly matches REQ-992, though timestamps align. |
Analysis
The 502 Bad Gateway error is a downstream timeout. Because the database entry exists with a timestamp preceding the error, the request reached the backend. Resending now would likely cause a duplicate entry or a 409 Conflict.
Next steps
- Do not resend REQ-992.
- Query the database for the specific record created at 10:05:01 UTC to verify the payload content.
- Update the local request status to 'Accepted' and close the reconciliation task.
Connects securely to your tools. The creator never sees your data.
About this skill
When a message, document, or API request enters a "pending" or "unknown" state, resending blindly often causes duplicate records or idempotent errors. This skill reconciles messy delivery evidence to determine the actual status of a previous submission without risking a re-transmission. It focuses on identifying specific submission fingerprints to separate successful deliveries from silent failures.
What it does
- Status reconciliation identifies whether a request was accepted, sent, delivered, or read based on available logs.
- Fingerprint matching uses targets, timestamps, and byte-level evidence to verify if a record belongs to the original intent.
- Gap analysis explicitly flags what evidence is missing and what cannot be attributed to the request.
- Idempotency protection prevents accidental re-submissions by strictly forbidding the replaying of intents.
- Conflict resolution handles diverging evidence, such as timeouts or post-acceptance errors, without assuming failure.
How it works
- Analyze input by reviewing the original request identifier, target, and content fingerprint.
- Scan observations for query times, entry points, and channel-specific status definitions.
- Determine conclusion by mapping evidence to a specific outcome like "accepted" or "outcome unknown."
- Draft reconciliation record including the basis for the conclusion and any necessary next steps.
Frameworks & tools
This skill is platform-agnostic and works with any messaging API, email log, or database audit trail. It is designed for environments where idempotency is not guaranteed at the transport layer.
Why this beats prompting it yourself
Standard prompts often hallucinate a "success" state when logs are ambiguous or suggest resending as a first step. This skill enforces strict boundary rules that prevent duplicate transactions and force a distinction between "not found" and "definitely not sent."
Use cases
- Reconciling stuck webhooks where the response was a 504 but the data might have landed.
- Auditing email delivery logs when a recipient claims non-receipt despite a "Sent" status.
- Investigating distributed system failures where a transaction ID appears in some logs but not others.
Known limitations
Does not perform active network queries or actual re-transmissions. It relies entirely on provided evidence and requires a byte-level fingerprint or specific identifier to guarantee identity.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Does not execute live API calls or re-sends. Requires existing log evidence and identifiers to function. Cannot resolve status if no logs or timestamps are provided.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 7 days ago
- Free with an Agensi account
Permissions
No special permissions declared or detected