Quickstart Skill

1---
2name: agentclash-quickstart
3description: Use when checking whether a workspace is ready to run AgentClash evals, interpreting quickstart readiness checks, or choosing the next CLI command after auth and workspace selection.
4metadata:
5  agentclash.role: onboarding
6  agentclash.version: "1"
7  agentclash.requires_cli: "true"
8---
9
10# AgentClash Quickstart
11
12## Purpose
13Run `agentclash quickstart` to verify auth, workspace, challenge packs, deployments, and baseline bookmark state, then report the CLI's suggested next command for eval or comparison workflows.
14
15## Use When
16- A user asks "am I ready to run an eval?" or "what should I run next?"
17- Auth and workspace are configured but pack/deployment readiness is unknown.
18- An agent needs a single read-only command before proposing `eval start` or `compare latest`.
19
20## Do Not Use When
21- CLI is not installed or auth failed — use `agentclash-cli-setup` first.
22- The user needs deep doctor diagnostics — use `agentclash doctor`.
23- The task is to create packs, deployments, or runs — use the eval-runner or challenge-pack skills after quickstart passes.
24
25## Inputs Needed
26- API URL (default hosted production).
27- Workspace ID or saved workspace from config / `AGENTCLASH_WORKSPACE`.
28- Whether the user wants structured JSON output for automation.
29
30## Environment
31```bash
32export AGENTCLASH_API_URL="https://api.agentclash.dev"
33agentclash auth login --device
34agentclash workspace use <WORKSPACE_ID>
35```
36
37## Procedure
381. Confirm auth with `agentclash auth status`.
392. Run `agentclash quickstart` (human) or `agentclash quickstart --json` (automation).
403. Read each check: `auth`, `workspace`, `challenge_packs`, `deployments`, `baseline`.
414. Execute `next_command` from the result when checks are blocking; treat `info` baseline as advisory.
425. When ready and baseline exists, quickstart suggests `agentclash eval start --follow` or comparison flows.
43
44## Commands
45```bash
46agentclash quickstart
47agentclash quickstart --json
48```
49
50Check names and statuses in structured output:
51
52| Check | ok | todo | info |
53| --- | --- | --- | --- |
54| `auth` | Token valid | Login required | — |
55| `workspace` | Workspace resolved | Set workspace | — |
56| `challenge_packs` | ≥1 pack visible | Init/publish pack | — |
57| `deployments` | ≥1 deployment | Create deployment | — |
58| `baseline` | Bookmark set | — | No baseline yet |
59
60Structured envelope fields:
61
62- `ready` — `true` when no blocking `todo` checks remain (`ok` and `info` are non-blocking).
63- `checks` — array of `{ name, status, detail, next_step?, metadata? }`.
64- `next_command` — first blocking `next_step`, or eval/compare suggestion when ready.
65- `next_steps` — ordered list including advisory baseline steps.
66
67When baseline is set and workspace is ready, quickstart may suggest:
68
69```bash
70agentclash eval start --follow
71agentclash compare latest --gate
72```
73
74## Expected Output
75Human output prints a bold **AgentClash Quickstart** header, per-check lines, and **Next Command** when applicable.
76
77JSON output prints the full envelope for scripting.
78
79## Failure Modes
80- `401` / auth errors → `agentclash auth login --device`.
81- No workspace → `agentclash link` or `agentclash workspace use <id>`.
82- Zero challenge packs → `agentclash challenge-pack init` then publish skills.
83- Zero deployments → `agentclash deployment create --from-file deployment.json`.
84- API unreachable → verify `AGENTCLASH_API_URL` and network.
85
86## Safety Notes
87- Quickstart is read-only; it does not create runs or spend provider budget.
88- Do not paste tokens from `--json` output into chat.
89
90## Report Back Format
91```text
92Quickstart ready: <yes/no>
93Checks:
94- auth: <status> — <detail>
95- workspace: <status> — <detail>
96- challenge_packs: <status> — <detail>
97- deployments: <status> — <detail>
98- baseline: <status> — <detail>
99Next command: <command or none>
100Next skill: <agentclash-eval-runner | agentclash-cli-setup | ...>
101```
102
103## Related Skills
104- `agentclash-hub`
105- `agentclash-cli-setup`
106- `agentclash-eval-runner`
107- `agentclash-compare-and-triage`
108
109## Related Docs
110- `/docs-md/getting-started/quickstart`
111- `/docs-md/getting-started/first-eval`
112- `/docs-md/reference/cli`