Resolve Overview

guided tour · part 1 of 5 resolve overview → connect an integration docid\ ig yfnpcmge2yt3oq5i4b → team knowledge docid\ d5xhezjtrrw k5spbvoob → auto investigations docid 3moxjnu2unbupnw evgda → tuning accuracy docid mempyfopnqpmkk 1vowo welcome to resolve this is the first stop in the guided tour a quick look at the main ways to interact with resolve so you know where everything lives before you start onboarding your own systems new to the product? read about resolve ai docid\ dovmei3t9gwkbtcm73uzp for the big picture first the main ways to work with resolve run alert investigations docid\ ctmycol8s0io8g1giztixchat with resolve about production docid\ ctmycol8s0io8g1giztixadd resolve to ongoing incidents docid\ ctmycol8s0io8g1giztixmonitor deployments and other production tasks docid\ ctmycol8s0io8g1giztixinteract with the resolve mcp server and rest api docid\ ctmycol8s0io8g1giztix 1\ run alert investigations open investigations from the sidebar you'll see two tabs at the top investigations — completed and in progress root cause analyses alerts — alerts flowing in from your monitoring tools from here you can kick off an investigation manually, or set alerts to auto investigate docid 3moxjnu2unbupnw evgda as they arrive you can filter by team to narrow the list open any investigation to see how resolve works a team of dynamic agents collaborate on the investigation the team scales up and down during the investigation as needed theories show the hypotheses the agents explored, which they converged on, and which they ruled out the evidence trail captures every query and data point the agents used the report (on the right) summarizes the investigation — typically with sections such as what happened , root cause , key evidence , and recommendations , though the exact structure adapts to the investigation hover any claim to see its citations , and open a citation in a new tab to dig into the underlying metrics or logs you can highlight text and ask follow up questions , @ mention a teammate to pull them in, and use the steer control on the left to redirect an investigation that went down the wrong path to learn how resolve decides how deep to go, see alert investigations docid\ z6j1w4ll2 pmc8te2syz resolve can also take action on its findings — see mitigation actions docid\ zlnb3zb3vezbe1qelhgbi for the full feature set, see features docid 53u1g03ypksugnlkwqliy 2\ chat with resolve about production chat is the fastest way to ask ad hoc questions ask things like "what services are running?" or "show me error logs from the payment service in the last hour" and resolve answers in plain english, drawing on the full context of your production systems — no query language required a chat is a lightweight conversation when a question needs deeper analysis, type /rca or click deep investigation in the chat window to escalate it into a full, multi agent deep investigation see alert investigation modes docid\ z6j1w4ll2 pmc8te2syz for how deep investigations work you can also chat wherever your team already works connect the app for slack docid 1h6snyqoiyvttj4utlq4v or app for ms teams docid\ qahtx0wbw9pc2 ba4clks and treat resolve like a teammate — ask "is there a bad deploy right now? i'm seeing errors in the front end" right in a channel and get a grounded, evidence backed answer see workplace collaboration tools docid\ suk9kssv8nrqvzkkiw6ap for how chat ties into team knowledge 3\ add resolve to ongoing incidents when an incident is underway, bring resolve into the channel where your team is already working resolve treats designated incident channels docid\ sun7tyfsahbp5azy9qcc as live investigations — it captures context from the conversation, links, and logs, then runs an autonomous investigation in the background and keeps it updated as the incident unfolds @ mention resolve to have it check logs, dashboards, deployments, or prs with full incident context catch up summaries privately bring new responders up to speed as they join (slack) postmortem drafts assemble a timeline, findings, impact, and supporting links when the incident winds down set this up with the app for slack docid 1h6snyqoiyvttj4utlq4v or app for ms teams docid\ qahtx0wbw9pc2 ba4clks , then see incident investigations docid\ sun7tyfsahbp5azy9qcc for the full workflow 4\ monitor deployments and other production tasks resolve can also work on autopilot tasks docid\ l mb8qpnclkejs8bfhkud let you tell resolve what to do, when to run, and where to report — using the same investigation abilities it brings to a live incident, but unattended on a schedule — hourly deploy checks, daily alert digests, weekly reports (interval or cron, in your time zone) on events — watch a slack or ms teams channel and act when a matching message appears (e g "watch #deploys for rollbacks and dm me if things trend poorly" ) on demand — one off checks that run now or at a due time you create tasks in plain language from a chat, a slack or ms teams thread, or the resolve ui, and each task reports back to a channel, a dm, or its own task page see tasks docid\ l mb8qpnclkejs8bfhkud for the full picture 5\ interact with the resolve mcp server and rest api for programmatic and agent to agent workflows, resolve exposes a resolve mcp server docid\ brcq0muxkokd9 cyxb1eb and a resolve rest api docid 84w41k1clcddxcqw3tihd install an ai assistant plugin docid\ bcqpoqdctungseqk8q5m to drive resolve from claude code , codex , or cursor — pull production context, join investigations, and turn findings into pull requests without leaving your editor connect any mcp client directly, or call the rest api docid 84w41k1clcddxcqw3tihd to trigger and query investigations from scripts and dashboards see the developers docid\ o vqioqmxfbd4rhujdhq section for everything you can build on top of resolve what's next now that you know the different ways to interact with resolve, the next step is connecting resolve to your systems so it can investigate real production data next part 2 — connect an integration docid\ ig yfnpcmge2yt3oq5i4b