docs: add DaoXE provider guide

[seven7763]

Issue

Fixes #12172

Context

DaoXE already exists in the models.dev catalog that Kilo refreshes, but users have no Kilo-specific setup page. This adds a focused guide for the existing provider path without adding provider code or a static model catalog.

The contributor is affiliated with DaoXE. This change was AI-assisted and cross-checked against Kilo's current provider source, the live models.dev entry, and the public DaoXE examples.

Implementation

The guide uses the existing daoxe provider ID and DAOXE_API_KEY metadata from models.dev. It documents VS Code and CLI setup, tells users to choose an account-available model ID, and limits the Kilo integration claim to the OpenAI-compatible chat path. The page also includes the service-region restriction and affiliation disclosure.

Screenshots / Video

Not applicable: this is a documentation-only page with no product UI change. The production docs build generated /ai-providers/daoxe successfully.

How to Test

Manual/local verification

• Agent-executed bun run --filter @kilocode/kilo-docs test: 4 test files and 22 tests passed.
• Agent-executed bun run --filter @kilocode/kilo-docs typecheck: passed.
• Agent-executed bun run --filter @kilocode/kilo-docs build: passed and generated the /ai-providers/daoxe static route.
• Verified the provider ID, environment variable, API base, and dynamic catalog behavior against the live models.dev data and Kilo's models-dev.ts implementation.

Reviewer test steps

1. Run bun run --filter @kilocode/kilo-docs dev.
2. Open /docs/ai-providers/daoxe.
3. Confirm the page appears under AI Gateways and that its internal and external links resolve.

Blocked checks and substitute verification

• No targeted docs checks are blocked.

Checklist

• Issue linked above, or exception explained
• Tests/verification described
• Screenshots/video included for visual changes, or marked N/A
• Changeset considered for user-facing changes; no changeset added because this only publishes documentation content
• I personally reviewed the diff and can explain the changes, including any AI-assisted work.

Get in Touch

GitHub discussion on this PR is the preferred contact path.

[vercel]

@seven7763 is attempting to deploy a commit to the Kilo Code Team on Vercel.

A member of the Team first needs to authorize it.

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

The incremental diff since 7530f15 is a single line: the intro sentence on packages/kilo-docs/pages/ai-providers/daoxe.md now describes DaoXE's differentiators (native Claude protocol, verifiable routing, one-key multi-model access) instead of the generic "multi-model API gateway" phrasing, which addresses the maintainer's request for a reason users would pick this provider. The daoxe provider ID and DAOXE_API_KEY references are unchanged and still match the live catalog metadata. Previously reported findings remain resolved on current code: both CLI references use kilo auth login --provider daoxe, and the VS Code tab correctly points at Kilo's auth.json credential store. No new issues.

Files Reviewed (1 files)

• packages/kilo-docs/pages/ai-providers/daoxe.md

Previous Review Summaries (5 snapshots, latest commit 7530f15)

Current summary above is authoritative. Previous snapshots are kept for context only.

Previous review (commit 7530f15)

Status: No Issues Found | Recommendation: Merge

The incremental diff since 8650ffa contains only the maintainer's suggested intro rewording on packages/kilo-docs/pages/ai-providers/daoxe.md (dropping the "single OpenAI-compatible endpoint" differentiator claim), plus an unrelated main merge. Previously reported findings are no longer present on current code: the CLI setup step and Troubleshooting entry both use kilo auth login --provider daoxe, and the VS Code tab correctly states credentials are stored in Kilo's auth.json. No new issues found.

Files Reviewed (1 files)

• packages/kilo-docs/pages/ai-providers/daoxe.md

Previous review (commit 8650ffa)

Status: No Issues Found | Recommendation: Merge

The latest commit (8650ffa) fixes both previously reported WARNING issues by replacing the invalid kilo auth daoxe invocation with kilo auth login --provider daoxe in the CLI setup step and the Troubleshooting section. No new issues found in this incremental diff.

Files Reviewed (1 files)

• packages/kilo-docs/pages/ai-providers/daoxe.md

Previous review (commit 1ee780f)

Status: 2 Issues Found | Recommendation: Address before merge

Overview

Severity	Count
CRITICAL	0
WARNING	2
SUGGESTION	0

Issue Details (click to expand)

WARNING

File	Line	Issue
packages/kilo-docs/pages/ai-providers/daoxe.md	42	kilo auth daoxe is not a real CLI invocation — the auth command only exposes list/login/logout; correct form is kilo auth login --provider daoxe
packages/kilo-docs/pages/ai-providers/daoxe.md	83	Same invalid kilo auth daoxe command repeated in Troubleshooting

The three prior reviewer comments (intro rationale, VS Code auth.json wording, and CLI /connect-first flow) were addressed in commit 1ee780fa1.

Files Reviewed (1 files)

• packages/kilo-docs/pages/ai-providers/daoxe.md

Fix these issues in Kilo Cloud

Previous review (commit 3f17f21)

Status: No Issues Found | Recommendation: Merge

Files Reviewed (1 files)

• packages/kilo-docs/pages/ai-providers/daoxe.md

Previous review (commit 5fabe0b)

Status: 1 Issues Found | Recommendation: Address before merge

Overview

Severity	Count
CRITICAL	0
WARNING	1
SUGGESTION	0

Issue Details (click to expand)

WARNING

File	Line	Issue
packages/kilo-docs/pages/ai-providers/daoxe.md	31	The VS Code flow stores provider credentials in auth.json, not kilo.json

Files Reviewed (3 files)

• packages/kilo-docs/lib/nav/ai-providers.ts - 0 issues
• packages/kilo-docs/pages/ai-providers/daoxe.md - 1 issue
• packages/kilo-docs/pages/ai-providers/index.md - 0 issues

Fix these issues in Kilo Cloud

---

_{Reviewed by claude-opus-5 · Input: 22 · Output: 3.9K · Cached: 431.5K}

_{Review guidance: REVIEW.md from base branch main}

[seven7763]

Addressed the review warning in 3f17f21: the VS Code section now correctly states that provider credentials are stored in Kilo’s auth.json, while kilo.json is identified as the separate CLI provider configuration. Re-ran the Kilo docs suite after the change: 22/22 docs tests, typecheck, and the production build all pass locally, including generation of /ai-providers/daoxe.

[seven7763]

Friendly bump: the review warning about auth.json vs kilo.json was addressed, and the Kilo review bot now recommends merge. Happy to adjust anything else.

[seven7763]

@kilocode-bot The VS Code credential path is corrected in the current docs: Providers UI → auth.json store; CLI remains DAOXE_API_KEY + kilo.json provider config only (no API key written into project kilo.json). Ready for another look.

[kilo-code-bot]

To use Kilo from GitHub you first need to link your GitHub account to Kilo. Link your Kilo account to continue. After linking, mention me again in this issue or pull request.

[lambertjosh]

Thanks - just a few minor comments to improve the docs for users.

[seven7763]

@lambertjosh Thanks — addressed in 1ee780fa1:

1. Intro — higher-level rationale (one multi-model OpenAI-compatible endpoint / one API key) instead of implementation-spec details.
2. VS Code — credential line is now: The provider credentials are stored in Kilo's auth.json store.
3. CLI — /connect (and kilo auth daoxe) is the recommended primary path so the key lands in auth.json; env + kilo.json kept as optional manual configuration.

Happy to tweak further if anything still feels off.

[seven7763]

@lambertjosh / @kilo-code-bot Follow-up fix in the latest commit:

• Replaced invalid kilo auth daoxe with kilo auth login --provider daoxe in the CLI tab and Troubleshooting (matches ProvidersCommand + other provider docs such as xAI).
• /connect remains the recommended interactive path so the key lands in auth.json.

Ready for another look.

[seven7763]

@lambertjosh All review threads are now resolved on the latest commits:

1. Intro is high-level (single multi-model OpenAI-compatible endpoint + one key), not implementation-spec detail.
2. VS Code credentials line: The provider credentials are stored in Kilo's auth.json store.
3. CLI primary path is /connect (and kilo auth login --provider daoxe); manual DAOXE_API_KEY + kilo.json is marked optional.
4. Invalid kilo auth daoxe references replaced with kilo auth login --provider daoxe (matches ProvidersCommand + other provider docs).

Kilo Code Review check is green. Ready for another look when you have a moment — no rush.

[seven7763]

@lambertjosh Quick check-in — all three review threads have been resolved in the latest commits (high-level intro, VS Code credentials path, CLI auth syntax). Is there anything else blocking re-approval? Happy to adjust further if needed.

[seven7763]

Thanks again for the review @lambertjosh! I believe all three points are now addressed in follow-ups 1ee780f and 8650ffa:

• Intro rationale — rewritten to a higher-level "why" (reach many models through one API key + base URL, i.e. a single OpenAI-compatible endpoint instead of wiring each vendor) rather than specifics.
• Credential storage — the VS Code tab now states credentials are stored in Kilo's auth.json store.
• CLI setup — the CLI tab now leads with /connect as the recommended path (key stored in auth.json, same as VS Code), with manual kilo.json + DAOXE_API_KEY kept as a secondary option. I also corrected the CLI command to kilo auth login --provider daoxe.

Would you mind taking another look when you have a moment? Happy to adjust anything further.

[seven7763]

Thanks @lambertjosh — all three notes make sense. I'll update the DaoXE provider page to:

1. Lead with higher-level rationale (why use this provider) instead of low-level specifics
2. State credentials live in Kilo's auth.json store
3. Prefer /connect (paste API key) as the primary path, with manual config as secondary

Also fixing the invalid kilo auth daoxe references if any remain (should be kilo auth login / UI /connect). Pushing a follow-up commit.

[seven7763]

Follow-up: the tip (docs(daoxe): fix CLI auth command… and current packages/kilo-docs/pages/ai-providers/daoxe.md) already reflects your review:

1. Higher-level rationale (single key / multi-model via one OpenAI-compatible endpoint) instead of low-level specs only
2. Credentials stored in Kilo's auth.json store
3. Primary path: TUI /connect (or kilo auth login --provider daoxe); manual kilo.json + env is secondary
4. Invalid kilo auth daoxe references replaced with kilo auth login --provider daoxe

Please re-review when you have a moment — happy to tweak wording further if anything still feels off.

[seven7763]

@lambertjosh gentle re-request: your three docs notes are in tip (1ee780f + 8650ffa):

1. Higher-level rationale (single key / multi-model OpenAI-compatible endpoint)
2. Credentials → Kilo auth.json
3. Primary path = TUI /connect (or kilo auth login --provider daoxe); manual config secondary

Vercel docs check still needs team authorization on your side (not fixable from the fork). Ready for re-review when you have a moment — happy to tweak wording further.

[seven7763]

@lambertjosh gentle ping — the 3 docs changes have been pushed. Let me know if anything else is needed.

[lambertjosh]

Set to auto-merge, thanks!