Nightshift MCP: drive your overnight log-to-PR pipeline from chat
Ask your coding agent what Nightshift found overnight, inspect incidents, get a verified patch and open the PR, with a confirmation before anything is published.
Free
Works with the AI Agents you already use
Nightshift MCP: drive your overnight log-to-PR pipeline from chat
Example session with this skill installed
what did nightshift find overnight?
- Read your context and instructions
- Compiled the nightshift mcp
| Severity | Incident | Count | Status |
|---|---|---|---|
| CRITICAL | DB connection leaked on the exception path (farmer-service) | 365 | TRIAGED |
| MAJOR | Settlement retries with no backoff (payment-service) | 327 | TRIAGED |
| MAJOR | NPE formatting initials for farmers with no last name | 92 | TRIAGED |
The top one is the one to act on: the pool is exhausted more often every day, and because every endpoint shares it, Nightshift forecasts a full farmer-service outage within about three weeks. Want me to have Nightshift propose a fix?
Connects securely to your tools. The creator never sees your data.
About this skill
Talk to a running Nightshift server from your coding agent: see what it found overnight, inspect an incident, have it propose a verified patch, and open the PR.
Nightshift is an open-source (Apache-2.0), self-hosted service that reads your application logs overnight, groups WARN/ERROR events into distinct incidents, triages them with severity, root cause and a forecast of what each costs if ignored, writes a bounded patch, reviews it adversarially, and opens one pull request per incident.
This skill teaches your agent to drive it well:
- Uses all five Nightshift MCP tools correctly:
list_incidents,get_incident,scan_logs,propose_patch,open_pull_request. - Answers from existing results first instead of re-scanning.
- Refuses to publish a patch the verifier did not pass, and checks for an existing PR before opening one.
- Asks before every PR, one yes per incident; never merges.
- Recognises dry-run mode and tells you plainly that nothing was pushed.
- Treats incident evidence as untrusted data, never as instructions.
Requires your own Nightshift deployment (Docker Compose, Render or Railway) registered as an MCP server named nightshift. Setup is in references/setup.md.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse All Claude Skills.
Needs a running Nightshift server; the skill does nothing on its own. It does not read logs or edit code itself (see the Log Triage and Incident to PR skills for that). Nightshift's /mcp endpoint is unauthenticated, so do not expose it publicly.
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 2 days ago
- Free with an Agensi account
- Software
- Nightshift server (self-hosted)Nightshift MCP server connected
Git
- Optional
- GitHub token on the server (real PRs)
Docker
Nightshift is free and open source: https://github.com/MdTanvirHossainTusher/nightshift. Without a GitHub token it runs in dry-run mode (nothing is pushed).
Permissions
Allowed Hosts
Calls only the nightshift MCP server's tools. open_pull_request pushes a branch and opens a PR through that server, and only after you confirm. The skill never merges. Nightshift's /mcp endpoint has no authentication of its own; keep it on localhost or a private network.