Commit Graph
4 Commits
Author SHA1 Message Date
Evan 38899238d7 chore: rename integrations/ directory to mcp/ (#64) 2026-04-17 23:38:44 +09:00
Evan 4f8dd2404d feat(providers): expand ollama catalog with thinking-capable models (#58)
* feat(providers): expand ollama model catalog with thinking-capable models

Add commonly used local models with accurate capability flags:
- gemma4, gemma3: supports_thinking, supports_vision
- deepseek-r1: supports_thinking (fix missing flag)
- deepseek-v3: supports_tools
- qwen3, qwq: supports_thinking
- llama4: supports_vision
- llama3.3: supports_tools
- phi4: supports_tools

Previously only 6 models were listed and none had supports_thinking
(except deepseek-r1), causing the dashboard to hide thinking toggles
for models that actually support it.

* chore(providers): major cleanup — remove defunct providers and old models

Delete 21 defunct/obscure providers:
aion-labs, arcee-ai, deepcogito, eleutherai, essentialai, ibm-granite,
inception, inflection, kwaipilot, lemonade, liquid, morph, nex-agi,
nousresearch, prime-intellect, reka, relace, switchpoint, tngtech,
upstage, writer

Clean up 10 major providers — keep only latest generation models:
- anthropic: remove claude-3.5-sonnet (superseded by 4.x)
- openai: remove gpt-4o/4-turbo/3.5/o1/o3-mini (superseded by gpt-5/4.1/o3/o4-mini)
- gemini: remove 1.5-*/2.0-flash (superseded by 2.5/3.x)
- deepseek: remove coder/chat-v3-0324 (superseded by r1/v3)
- qwen: remove turbo/2.5-coder (superseded by qwen3)
- groq: remove old llama/mixtral/gemma (keep latest only)
- mistral: remove medium/nemo/pixtral-large (keep large/small/codestral)
- xai: remove grok-2 (superseded by grok-3/4)
- meta-llama: remove 3.x/guard (keep llama-4 + 3.3)
- ollama: rewrite with current models (gemma4, qwen3, qwq, llama4, etc)

Total: 90 → 48 models across major providers. All thinking-capable
models now have supports_thinking = true.

* chore: add pre-commit hook for automatic TOML formatting

- .githooks/pre-commit: runs taplo fmt on staged .toml files
- Makefile: add setup target + auto-configure hooks on first make
- .gitignore: add .make-setup-done and .sync_marker
2026-04-15 22:18:01 +09:00
Evan d1cab3e33a feat: context engine plugins, scaffolding, and pricing fixes (#6)
* feat: add 4 context engine plugins

- topic-memory: keyword clustering for topic-aware memory recall
- episodic-memory: conversation segmentation and cross-session recall
- user-profile: persistent user profiling from conversation patterns
- context-decay: time-based memory decay with reinforcement dynamics

All plugins use the ingest/after_turn hook protocol with stdin/stdout JSON.

* chore: add plugin scaffolding, update docs and templates

- Add plugin.toml template with {{NAME}} placeholder
- Add new-plugin Makefile target with hooks/ scaffolding
- Update plugins/README.md with all 10 plugins
- Update README.md stats (10 plugins, 220+ models)
- Add Plugin checkbox and checklist to PR template
- Add Plugin to issue template content type dropdown
- Fix CONTRIBUTING.md: last_verified is recommended, not required

* fix: correct model pricing and remove deprecated entries

- openrouter/gemma-2-9b-it: fix pricing from 0.0 to 0.03/0.09 per M tokens
  (free variant correctly stays at 0.0)
- github-copilot: remove deprecated copilot/gpt-4 model entry
  (GPT-4 retired in favor of GPT-4o for Copilot)

* docs: annotate kimi-coding as membership-gated

Kimi Code CLI uses quota-based membership model (not per-token billing).
Free tier has limited weekly requests; underlying model is K2.5.
Pricing kept at 0.0 consistent with other subscription providers
(chatgpt, github-copilot) but with explanatory comments.

* style: fix trailing newline in github-copilot.toml

* fix: correct Moonshot/Kimi model pricing from official sources

All 5 models had incorrect pricing:
- moonshot-v1-8k: 0.10/0.10 → 0.20/2.00
- moonshot-v1-32k: 0.30/0.30 → 1.00/3.00
- moonshot-v1-128k: 0.80/0.80 → 2.00/5.00
- kimi-k2: 2.00/8.00 → 0.60/2.50
- kimi-k2.5: 2.00/8.00 → 0.45/2.20

Sources: platform.moonshot.ai/docs/pricing/chat, costgoat.com, getmaxim.ai

* feat: add MiniMax M2.7 and M2.7-highspeed models

Released 2026-03-18, MiniMax's latest flagship text model.
10B activated params, 200K context, 128K output, tool use, streaming.
Pricing: $0.30/$1.20 per M tokens (input/output).

Added to both international (minimax.io) and China (minimaxi.com) providers.
2026-03-21 03:36:32 +09:00
Evan Hu a8b7c9d08b feat: comprehensive registry improvements
Community docs:
- CODE_OF_CONDUCT.md (Contributor Covenant v2.1)
- SECURITY.md (vulnerability reporting policy)
- CHANGELOG.md (initial release notes)
- CODEOWNERS (per-type review ownership)

GitHub config:
- Issue templates: bug-report, pricing-correction, documentation
- FUNDING.yml (GitHub Sponsors)

Validation enhancements:
- Cross-reference check: hand [[requires]] → integration existence
- Routing alias collisions as warnings (errors with --strict)
- --strict flag to promote warnings to errors
- --type filter to validate single content type
- CI: add taplo format check and lychee link check jobs

Developer experience:
- Makefile with validate, fmt, and scaffold targets
- Scaffold templates for all 5 content types
- .pre-commit-config.yaml (trailing whitespace, TOML check, validate)
- docs/content-guide.md (naming, descriptions, prompts, decision guide)

Content quality:
- schema.toml: add last_verified field for model pricing
- CONTRIBUTING.md: add pricing verification guide with source links
2026-03-21 02:46:04 +09:00