alex/club-3090
1
0
Fork 0
Code Issues Pull Requests Actions 32 Packages Projects Releases Wiki Activity

docs: link the docs index + ARCHITECTURE from the repo agent entry points

Browse Source 
AGENTS.md 'Read first' + the new CLAUDE.md now point at docs/README.md (the
canonical user/contributor docs index) + ARCHITECTURE.md/GETTING_STARTED/
BENCHMARKS/FAQ/KV_MATH, so a user's AI agent landing on either entry point can
navigate the whole doc set instead of just the add-a-model path.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
noonghunna
2026-05-26 22:19:49 +00:00
parent 37871c6317
commit dc7c613262
2 changed files with 10 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
AGENTS.md
View File

@@ -7,7 +7,10 @@ Guidance for AI coding agents (Claude Code, Cursor, Copilot, Continue, etc.) wor
Before making non-trivial changes: Before making non-trivial changes:
- [`README.md`](README.md) — what the repo is, two-routes framing, repo layout - [`README.md`](README.md) — what the repo is, two-routes framing, repo layout
- [`docs/README.md`](docs/README.md) — **the docs index** (user track + contributor track); start here to find any guide
- [`docs/ARCHITECTURE.md`](docs/ARCHITECTURE.md) — current stack state (services, ports, paths, model + KV config)
- [`docs/SINGLE_CARD.md`](docs/SINGLE_CARD.md) / [`docs/DUAL_CARD.md`](docs/DUAL_CARD.md) — pick-by-workload guidance + the cliffs they reference - [`docs/SINGLE_CARD.md`](docs/SINGLE_CARD.md) / [`docs/DUAL_CARD.md`](docs/DUAL_CARD.md) — pick-by-workload guidance + the cliffs they reference
- [`docs/ADDING_MODELS.md`](docs/ADDING_MODELS.md) — add a model: serve-locally vs the curated-catalog workflow (see "Adding a model" below)
- [`docs/UPSTREAM.md`](docs/UPSTREAM.md) — every upstream issue / PR we depend on or have filed (see "Upstream issues" section below for why this matters) - [`docs/UPSTREAM.md`](docs/UPSTREAM.md) — every upstream issue / PR we depend on or have filed (see "Upstream issues" section below for why this matters)
- [`models/qwen3.6-27b/INTERNALS.md`](models/qwen3.6-27b/INTERNALS.md) — DFlash forensics, AutoRound rationale, Marlin pad fork, MTP head - [`models/qwen3.6-27b/INTERNALS.md`](models/qwen3.6-27b/INTERNALS.md) — DFlash forensics, AutoRound rationale, Marlin pad fork, MTP head
- [`CONTRIBUTING.md`](CONTRIBUTING.md) — what kinds of PRs land cleanly + benchmark + verify protocol - [`CONTRIBUTING.md`](CONTRIBUTING.md) — what kinds of PRs land cleanly + benchmark + verify protocol

7
CLAUDE.md
View File

@@ -6,6 +6,13 @@ first. It's the canonical, engine-agnostic guide: conventions, the compose
single-source-of-truth + `DEFAULTS` resolver, hardware truths, the test pipeline, single-source-of-truth + `DEFAULTS` resolver, hardware truths, the test pipeline,
and what **not** to do. and what **not** to do.
**Find a doc:** [`docs/README.md`](docs/README.md) is the full index (user track +
contributor track — links every guide). Quick anchors:
[`docs/GETTING_STARTED.md`](docs/GETTING_STARTED.md) (clone-to-curl) ·
[`docs/ARCHITECTURE.md`](docs/ARCHITECTURE.md) (current stack state) ·
[`BENCHMARKS.md`](BENCHMARKS.md) (measured numbers) ·
[`docs/FAQ.md`](docs/FAQ.md) · [`docs/KV_MATH.md`](docs/KV_MATH.md).
**Adding a model?** Start at AGENTS.md → "Adding a model", then follow the full **Adding a model?** Start at AGENTS.md → "Adding a model", then follow the full
workflow in **[docs/ADDING_MODELS.md](docs/ADDING_MODELS.md)** (three paths: workflow in **[docs/ADDING_MODELS.md](docs/ADDING_MODELS.md)** (three paths:
serve a safetensors repo via `pull.sh`, run a local GGUF, or promote into the serve a safetensors repo via `pull.sh`, run a local GGUF, or promote into the