# Analyze

> AI-summarise a failed Dokploy deployment or a crashing container via the configured ai provider Arguments: deployment-id|app-name-or-id|compose-name-or-id.

- Skill: `agents-store/analyze` (Agent Skill)
- Install (CLI): `npx get-skill add agents-store/analyze`
- Raw SKILL.md: https://api.skillmd.com/api/skills/agents-store/analyze/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: agents-store (https://skillmd.com/u/agents-store)
- Updated: 2026-11-16
- Page: https://skillmd.com/skills/agents-store/analyze

---


# AI-Analyse a Failure

Pipeline around `ai-analyzeLogs`: fetch the relevant log text, then ask the configured AI provider for a root cause + suggested fix. `ai-analyzeLogs` takes the **log text** (not a `deploymentId`) plus an `aiId` and a `context`.

Follow the `ai-assist` skill (`${CLAUDE_PLUGIN_ROOT}/skills/ai-assist/SKILL.md`), Step 3, for the analyze step. The log text comes from the `read-logs` skill (§1 runtime logs, §3 build logs, §5 hand-off to AI triage).

## Arguments

Format: `[deployment-id|app-name-or-id|compose-name-or-id]` (optional)

- Deployment id → analyse that build log (`context: "build"`).
- App / compose name or id → fetch its current runtime log (`context: "runtime"`); for a failed build, fetch the most recent errored deployment's build log instead.
- No argument → `deployment-allCentralized`, list recent `status: error` rows, ask which to analyse.

Parse from "$ARGUMENTS".

## Process

1. **Verify AI is configured** — `mcp__plugin_dokploy-dev_dokploy__ai-getEnabledProviders`.
   - If the array is **empty**: tell the user no AI provider is enabled, offer `/dokploy-dev:debug <id>` (manual decision tree), and point at the `ai-assist` skill to wire one up. Stop.
   - Else: take the `aiId` of an enabled provider.

2. **Fetch the log text** (this is the `logs` argument):
   - Build failure → `deployment-readLogs { deploymentId, tail: 1000 }` → `context: "build"`.
   - App runtime → `application-readLogs { applicationId, tail: 500, since: "2h" }` → `context: "runtime"`.
   - Compose runtime → enumerate containers and concatenate each container's `compose-readLogs { composeId, containerId, tail }` (or run `/dokploy-dev:compose-logs` first) → `context: "runtime"`.
   - Truncate very large logs before sending (provider context limits).

3. **Run the analysis:**

   ```
   mcp__plugin_dokploy-dev_dokploy__ai-analyzeLogs
     → { aiId: "<enabled provider id>", logs: "<text from step 2>", context: "build" | "runtime" }
   ```

4. **Present the result:**

   ```
   ## AI Analysis: <resource name>

   **Context:** build | runtime
   **Provider:** <providerName>/<model>

   ### Root cause
   <ai-analyzeLogs root cause>

   ### Suggested fix
   <ai-analyzeLogs suggested fix>

   ### Verify before applying
   - Cross-check against the raw log (`/dokploy-dev:logs <id>` or `/dokploy-dev:compose-logs <name>`).
   - The model only sees the log you sent it, not the codebase — trust the log if they conflict.
   ```

5. **Offer follow-up:**
   - "Apply this fix? I can update env / build type / Dockerfile and redeploy."
   - "Want the full decision tree? Run `/dokploy-dev:debug <id>`."

## Example Usage

```
/dokploy-dev:analyze
/dokploy-dev:analyze web-frontend
/dokploy-dev:analyze my-compose-stack
/dokploy-dev:analyze <deploymentId>
```
