Ship docs in the package, fix the site path in the README

[DavertMik]

Found while checking the explorbot-setup skill against docs/workflow/agentic-usage.md. Docs only, plus one line of package.json.

docs/ ships in the npm package

files listed dist/, src/, bin/, rules/, assets/sample-files/, and models.json — no docs. So node_modules/explorbot/docs/ did not exist on any install, and an agent asking "what does --configure do" had nothing local to read. Confirmed against a real install.

Doc fixes

• README.md said output lands in ~/.explorbot/state/<host>/; sites are registered under ~/.explorbot/sites/<host>/.
• agentic-usage.md and commands.md now state that the global config holds models and keys and never a site, and that the commands taking no URL argument — test, learn, knows, experience, compact — read EXPLORBOT_URL. That was undocumented, and the failure (No site to explore) is easy to hit on a first agentic run.
• commands.md gains the missing mistral in its provider list.

An earlier revision of this PR added init --global --url to pin a site in the global config. Dropped: the global install is machine-wide and every run names its own site, so a default site there is the wrong shape.

🤖 Generated with Claude Code

[github-actions]

Explorbot Self-Regression

Commit f453819 · run

Scenario	Result	Attempts	Duration
basic (native)	PASS	1/3	6m
experience: control	OK — failed as expected	1/1	33s
experience: seeded	PASS	1/3	54s

Attempt details

• basic (native) attempt 1 — PASS: login evidence: PASS (post-login plan=true, post-login research=true); research: PASS (files=7, wellFormed=true, keywords=7/3); scenarios: PASS (tests=7/5, features=4/3); tests passed: PASS (5 passed, 0 failed (reporter: 5 passed, 0 failed))
• experience: control attempt 1 — PASS: control: OK — failed as expected (0 passed, 1 failed)
• experience: seeded attempt 1 — PASS: seeded: PASS (1 passed, 0 failed)

---

Session analysis — basic (native):

Session Analysis

The Issues module core functionality was validated: creating issues, searching, filtering by label/status, and viewing details all work as expected. Multiple automation retry patterns observed across tests but all passed verification.

Coverage

• Pages: /issues, /issues/<id>
• Features: Issue creation, keyword search, label filtering, status filtering, detail view

What works

• Issue creation — ET-1 Create new issue and verify in list
• Keyword search — ET-2 Search for existing issue by keyword
• Label filtering — ET-3 Filter issues by label using dropdown
• Status filtering — ET-4 Filter issues by status and verify results
• Detail view — ET-5 View issue details by clicking issue link

Execution Issues

• ET-1 — Form fill required 2 attempts before submit succeeded
• ET-3 — Label dropdown click failed 4 times before filter applied
• ET-4 — Status option click failed 3 times before filter applied