# signals-scout-tasks

signals-scout-tasks — a scout skill for PostHog Tasks: it tracks how task runs get delivered and, on a slower pass, looks for the same request coming up again in tasks people wrote.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode, Windsurf (Devin Desktop), Claude.ai · 179 installs · ★ 101 repo stars
- Author: PostHog
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/PostHog/ai-plugin/tree/HEAD/skills/signals-scout-tasks/SKILL.md
- Updated: 2026-11-16
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/skills/signals-scout-tasks

## What it does

The skill runs inside the PostHog Signals agent in a Claude sandbox and looks at a project's tasks: what people asked for, and how their runs ended. It reads Postgres system tables `system.tasks` and `system.task_runs` with `execute-sql`. Findings go out through `emit_report` and `edit_report`, and only when they clear a bar.

That bar is failed runs divided by the number of distinct tasks that failed. Near 1 across many tasks points to one defect on a shared path, the best thing this lens can find. Far above 1 on a task or two points to something stuck and being retried. A repository at total failure is reported only when at least two different tasks broke. `error_message` shows up on nearly every failed run, so it proves nothing without a `status = 'failed'` filter. Probe `task_runs.stage` with `countIf(stage != '')` before relying on it.

Nothing runs without the MCP scopes `task:read`, `signal_scout_internal:write` and `signal_scout_report:write`, plus `execute-sql`, `tasks-retrieve` and the inbox tools. Always drop `origin_product` values `signals_scout` and `scout_suggestions`: those rows are the scout's own run containers, they can carry the most rows on a busy project, and a report that keeps them grades itself. To name a reviewer, take `created_by_id` (an integer with no lookup) through `tasks-retrieve` and read `created_by.uuid`. The SQL lives in `references/queries.md`.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add PostHog/ai-plugin --skill signals-scout-tasks -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/signals-scout-tasks/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/signals-scout-tasks/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add PostHog/ai-plugin --skill signals-scout-tasks -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/signals-scout-tasks/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add PostHog/ai-plugin --skill signals-scout-tasks -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/signals-scout-tasks/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install PostHog/ai-plugin skills/signals-scout-tasks
```

2. **Manually: put the files in place**
- File: `.github/skills/signals-scout-tasks/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add PostHog/ai-plugin --skill signals-scout-tasks -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/PostHog/ai-plugin --path skills/signals-scout-tasks
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/signals-scout-tasks/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add PostHog/ai-plugin --skill signals-scout-tasks -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/PostHog/ai-plugin/tree/HEAD/skills/signals-scout-tasks/SKILL.md

## When to use it
- spot failures piled into one repository instead of spread across the project
- tell a shared defect apart from one task being retried
- find repeated requests in human-written tasks
- keep error messages on cancelled runs from reading as failures
- filter the scout's own runs out of the sample

## Example prompts
- Run the tasks scout and show me where failures sit in one repository
- Check whether people keep asking for the same thing in their tasks
- Pull a delivery report for the last 14 days

## Similar
- [signals-scout-web-analytics](https://affpapa.org/en/nevibecode/skills/signals-scout-web-analytics.md) — skill · 317 installs
- [signals-scout-replay-vision](https://affpapa.org/en/nevibecode/skills/signals-scout-replay-vision.md) — skill · 316 installs
- [speckit-tasks](https://affpapa.org/en/nevibecode/skills/speckit-tasks.md) — skill · 2500 installs
- [capture-tasks-from-meeting-notes](https://affpapa.org/en/nevibecode/skills/capture-tasks-from-meeting-notes.md) — skill · 1 install
- [brief-to-tasks](https://affpapa.org/en/nevibecode/skills/brief-to-tasks.md) — skill · 5400 installs
- [c-tasks](https://affpapa.org/en/nevibecode/skills/c-tasks.md) — skill · ★ 179 repo stars

From the same set: [posthog/ai-plugin](https://affpapa.org/en/nevibecode/by/posthog/ai-plugin.md) — 82 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
