# Truto CLI Toolbelt

> Shared Truto CLI operating playbook for profiles, command discovery, JSON output, account data-plane calls, logs, safe writes, and common gotchas. Use whenever an agent will run Truto CLI commands or prepare another Truto debugging skill.

- Skill: `yuvraj3335/truto-cli-toolbelt` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add yuvraj3335/truto-cli-toolbelt`
- Raw SKILL.md: https://api.skillmd.com/api/skills/yuvraj3335/truto-cli-toolbelt/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Product & Planning
- Author: yuvraj3335 (https://skillmd.com/u/yuvraj3335)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/yuvraj3335/truto-cli-toolbelt

---


# Truto CLI Toolbelt

## Intent

Use this skill as the baseline for Truto CLI work. The CLI is an admin and debugging surface for Truto platform state, not application code. It manages integrations, environment integrations, accounts, docs, sync jobs, workflows, webhooks, logs, unified/proxy/custom APIs, exports, diffs, and mappings.

## Fast Start

For short mobile prompts, infer what you can and start read-only:

```bash
truto whoami -p "$PROFILE" -o json --no-color
truto context | sed -n '1,220p'
```

If no profile is provided, try the active profile with `truto whoami -o json --no-color`. Ask for a profile or token only if the command fails or the environment is ambiguous.

## Hard Rules

- Use `-o json` for structured admin calls and `-o ndjson` for streaming exports.
- Use `--no-color` for outputs you will parse or quote.
- Use `-v` only to debug HTTP behavior. Redact bearer tokens, cookies, API keys, OAuth secrets, authorization headers, and customer PII before reporting.
- Start with read-only inspection. Treat create/update/delete, override helpers, refresh credentials, test deliveries, schedules, sync-job-run creation, file uploads, and provider write methods as mutating.
- Remember that the API token is environment-scoped. Most commands do not need an `environment_id`; profile choice determines the environment.
- If command shape is uncertain, run `truto <command> --help`, `truto <command> <subcommand> --help`, or `truto context --full` before guessing.

## Entity Ladder

Resolve Truto objects in this order:

```bash
truto accounts get "$ACCOUNT_ID" -p "$PROFILE" -o json --no-color
truto accounts tools "$ACCOUNT_ID" -p "$PROFILE" -o json --no-color
truto capabilities "$ACCOUNT_ID" --target account -p "$PROFILE" -o json --no-color
truto environment-integrations get "$ENV_INTEGRATION_ID" -p "$PROFILE" -o json --no-color
truto environment-integrations show-override "$ENV_INTEGRATION_ID" -p "$PROFILE" -o json --no-color
truto integrations get "$INTEGRATION_ID" -p "$PROFILE" -o json --no-color
```

If only an integration slug is known:

```bash
truto integrations list --name "$INTEGRATION" -p "$PROFILE" -o json --no-color
truto capabilities "$INTEGRATION" --target integration -p "$PROFILE" -o json --no-color
```

## Data-Plane Choice

- Use `unified` when the question is about normalized schemas and cross-integration behavior:
  ```bash
  truto unified "$MODEL" "$RESOURCE" -m list -a "$ACCOUNT_ID" -p "$PROFILE" -o json --no-color
  ```
- Use `proxy` when the question is about provider-native resources or integration config:
  ```bash
  truto proxy "$RESOURCE" -m list -a "$ACCOUNT_ID" -p "$PROFILE" -o json --no-color
  ```
- Use `custom` when the provider endpoint exists but is not mapped:
  ```bash
  truto custom "$PATH" -m GET -a "$ACCOUNT_ID" -p "$PROFILE" -o json --no-color
  ```
- Prefer `--query-json` for nested filters and `--stdin` for large bodies.

## Evidence Standard

Do not stop at "it failed." Collect enough evidence to identify the layer:

- CLI scope: profile, API URL, team/environment from `whoami`
- Target: account, environment integration, catalog integration, model/mapping/docs when relevant
- Capability: `accounts tools` and `capabilities`
- Logs: `unified_proxy_api`, `rapid_bridge`, `webhook`, `sync_job_cron_trigger`, or `mcp`
- Reproduction: smallest read-safe command, then verbose only if needed

## Final Answer Contract

Return: profile used, target IDs, commands run, key evidence, likely layer, confidence, and next safe action. Never include raw secrets or full noisy payloads unless the user explicitly asks.