CLI Finder

Task playbooks

CLI Workflows for AI Agents

Start from an outcome such as reviewing a pull request, querying a database, or deploying a Worker.

Choose the playbook that matches the target system and acceptable write risk.

Recommended starting points

Start with a common decision or workflow, then narrow the commands and permissions for your environment.

Deployment and cloudRemote write

Deploy a Next.js application

Build locally, create a preview deployment, verify the result, and keep production promotion behind an explicit approval.

Produce a working preview and a release-ready report without silently changing production.

4 tools3 approval points
Open the complete guide
Code collaborationRead only

Review GitHub pull requests

Collect PR metadata, inspect the patch locally, and report blockers before posting or merging anything.

Produce an evidence-backed review that separates findings from repository-changing actions.

3 tools3 approval points
Open the complete guide
Code collaborationRemote write

Triage GitHub issues

Read open issues as structured data, group them by evidence and urgency, then propose rather than silently apply changes.

Turn an issue queue into an actionable triage report without losing user context or changing state unexpectedly.

2 tools3 approval points
Open the complete guide

CLI workflows for AI agents by task

Each destination is a complete, source-backed page with a clear outcome and operating boundary.

Deployment and cloud

Deployment and cloudRemote write

Deploy a Next.js application

Build locally, create a preview deployment, verify the result, and keep production promotion behind an explicit approval.

Produce a working preview and a release-ready report without silently changing production.

4 tools3 approval points
Open the complete guide
Deployment and cloudRemote write

Deploy a Cloudflare Worker

Validate configuration and build output, run a local or preview check, then deploy only to the named environment and verify it.

Publish a Worker to the intended account and environment with a known rollback target.

3 tools3 approval points
Open the complete guide

Code collaboration

Code collaborationRead only

Review GitHub pull requests

Collect PR metadata, inspect the patch locally, and report blockers before posting or merging anything.

Produce an evidence-backed review that separates findings from repository-changing actions.

3 tools3 approval points
Open the complete guide
Code collaborationRemote write

Triage GitHub issues

Read open issues as structured data, group them by evidence and urgency, then propose rather than silently apply changes.

Turn an issue queue into an actionable triage report without losing user context or changing state unexpectedly.

2 tools3 approval points
Open the complete guide

Containers and Kubernetes

Containers and KubernetesRead only

Read-only Kubernetes audit

Pin the context and namespace, use get, describe, logs, and auth checks, and reject mutating subcommands.

Explain cluster health and likely causes while keeping the audit technically read-only.

2 tools3 approval points
Open the complete guide

Databases

DatabasesRead only

Query PostgreSQL

Use psql non-interactively with a read-only role, bounded queries, and machine-readable output.

Answer a database question without exposing credentials or modifying data.

2 tools3 approval points
Open the complete guide
DatabasesRead only

Query Supabase

Inspect local project state first, identify the linked project explicitly, and use a read-only database path for production data.

Answer Supabase project or database questions without crossing project boundaries or bypassing data protections.

2 tools3 approval points
Open the complete guide

HTTP and webhooks

HTTP and webhooksRemote write

Test Stripe webhooks

Forward test events to a local endpoint, trigger a known fixture, and verify signature handling and idempotency.

Prove a webhook handler accepts and processes test events without touching live customers or money.

1 tools3 approval points
Open the complete guide
HTTP and webhooksLocal write

Debug an HTTP API

Reproduce one request with explicit method, headers, body, timeout, and saved response before changing the server or client.

Identify whether a failure comes from transport, authentication, request shape, server behavior, or response parsing.

3 tools3 approval points
Open the complete guide
HTTP and webhooksLocal write

Extract content from a website

Define a small URL scope, choose extraction or browser automation, and preserve each source URL with the result.

Collect the specific public content needed for a task without uncontrolled crawling or hidden interaction.

2 tools3 approval points
Open the complete guide

Data and text

Data and textLocal write

Transform JSON

Inspect the input shape, build a deterministic jq filter, validate the result, and only then replace a file if needed.

Produce valid JSON with the requested schema while preserving the original input.

1 tools3 approval points
Open the complete guide
Data and textLocal write

Analyze CSV files

Profile the file, query it locally with DuckDB or csvkit, and save the analysis separately from the source.

Answer a tabular data question with reproducible commands and clearly stated data-quality limits.

3 tools3 approval points
Open the complete guide

Operations and infrastructure

Operations and infrastructureLocal write

Search application logs

Constrain the time, service, file, and pattern first; then collect only the lines needed to explain the incident.

Find a relevant error sequence without exposing unrelated users, secrets, or an unbounded production log stream.

3 tools3 approval points
Open the complete guide

How to choose the right guide

Start from the outcome

Choose the page that names the system and result you need, not only a familiar tool.

Check the highest risk

Read-only work, local edits, remote writes, and destructive actions need different review levels.

Match the identity

Confirm which account, token, cluster context, or database role the command will use.

Keep a recovery path

Prefer workflows that separate preview, verification, approval, and rollback.

What every guide gives you

Inputs and outputs

Know what context the agent needs and what evidence it must return.

Copyable steps

Run bounded commands with explicit risk labels and expected results.

Approval and rollback

See exactly where a person decides and how to recover from an approved write.

Interface alternatives

Decide when the same task fits a CLI, MCP integration, or direct API.

Continue through the registry

Evidence-labeled CLI tools

Inspect commands, evidence, output formats, and risk before using a tool.

Open guide

Agent CLI stacks

Adapt a workflow to your coding agent and its approval model.

Open guide

Interface comparisons

Choose between CLI, MCP, and direct APIs for a specific operation.

Open guide