Commit graph

5 commits

Author SHA1 Message Date
Florian BRUNIAUX
18ea240e12 docs: add MCP Apps (SEP-1865) documentation
Integrate comprehensive documentation for MCP Apps, the first official
MCP extension enabling interactive UI delivery.

Changes:
- guide/architecture.md (656): New section "MCP Extensions: Apps"
  - Technical architecture (primitives, SDK, security)
  - Platform support (Claude Desktop, VS Code, ChatGPT, Goose)
  - Example implementations (9 production tools at launch)
  - Developer workflow and SDK usage
  - ~150 lines of technical documentation

- guide/ultimate-guide.md (6509): New section "MCP Evolution: Apps"
  - User context and use cases
  - Available interactive tools (Asana, Slack, Figma, etc.)
  - Platform support matrix
  - Hybrid workflow examples
  - ~90 lines of user-facing documentation

- guide/ultimate-guide.md (7522): Table update
  - Added "Interactive UI" row to Plugin vs. MCP Server comparison
  - Clarified MCP Apps = "What Claude can show"

- machine-readable/reference.yaml: 8 new entries
  - mcp_apps_architecture, mcp_apps_evolution
  - Links to spec, SDK, blog posts
  - CLI relevance note (indirect)

- docs/resource-evaluations/mcp-apps-announcement.md: New evaluation
  - Score: 4/5 (High Value - Integrate within 1 week)
  - Fact-checked with Perplexity searches
  - Technical review by agent

Resource evaluated:
- https://blog.modelcontextprotocol.io/posts/2026-01-26-mcp-apps/
- https://claude.com/blog/interactive-tools-in-claude

Total documentation: ~240 lines across 3 files
Score: 4/5 (High Value)
CLI relevance: Indirect (ecosystem understanding, MCP server dev, hybrid workflows)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-27 08:14:49 +01:00
Florian BRUNIAUX
3a7671ac5e docs: update $ARGUMENTS syntax for v2.1.19 breaking change + evaluation
Updated all documentation and examples to reflect Claude Code v2.1.19
breaking change: $ARGUMENTS.0 → $ARGUMENTS[0] (bracket syntax).

Changes:
- guide/ultimate-guide.md: 7 occurrences updated to bracket/shorthand syntax
- guide/cheatsheet.md: Command template updated
- Added migration note in § 6.2 Variable Interpolation
- Created migration scripts: migrate-arguments-syntax.{sh,ps1}
  • Automated detection + conversion with backups
  • Dry-run mode, cross-platform (macOS/Linux/Windows)
- Added formal evaluation: eval-claude-code-releases-jan2026.md
  • Score: 5/5 (Critical - Integrate Immediately)
  • Covers releases 2.1.0 to 2.1.19 (January 2026)
  • Technical accuracy verified against GitHub CHANGELOG

Files:
- guide/ultimate-guide.md (+23 lines, 7 occurrences fixed)
- guide/cheatsheet.md (+1 line)
- examples/scripts/migrate-arguments-syntax.sh (+152 lines)
- examples/scripts/migrate-arguments-syntax.ps1 (+143 lines)
- docs/resource-evaluations/eval-claude-code-releases-jan2026.md (+273 lines)
- CHANGELOG.md (+12 lines, Unreleased section)

Total: +605 lines

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-26 17:37:49 +01:00
Florian BRUNIAUX
08a2a4261f docs: add system prompts official sources (Anthropic + community analyses)
- Add System Prompt Contents section in architecture.md (line 300-326)
  - Official Anthropic sources (Tier 1, 100% confidence)
  - Community analyses (Simon Willison, PromptHub)
  - Claude.ai vs API vs Code CLI distinctions

- Update reference.yaml with 4 new system prompts entries
  - system_prompts_official, willison_analysis, prompthub, architecture

- Create resource evaluation: system-prompts-official-vs-community.md
  - Score evolution: 4/5 → 3/5 → 2/5 (after Perplexity fact-check)
  - Decision: Do not integrate x1xhlol repository (redundant)
  - Use official Anthropic sources instead

- Create watchlist for non-integrated resources
  - claudedocs/resource-evaluations/watch-list.md (gitignored)
  - Re-evaluation triggers: Q2 2026 or if 10+ user requests

Related evaluation: docs/resource-evaluations/system-prompts-official-vs-community.md

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-26 17:03:42 +01:00
Florian BRUNIAUX
ab8bfcf782 docs: add Anaconda Croce evaluation (minimal integration)
Resource evaluated: "What I Learned Challenging Claude to a Coding Competition"
by Steve Croce (Anaconda Field CTO, Jan 16, 2026)

Score: 2/5 (Marginal - Info secondaire)

Integration:
- Added "Community Experiences" section in guide/learning-with-ai.md
- 2-paragraph mention with strong caveats (N=1, non-representative context)
- Full evaluation in docs/resource-evaluations/anaconda-croce-evaluation.md
- Updated reference.yaml count (14 → 16 evaluations)

Rationale:
- Provides light empirical validation (90s vs 60min on Advent of Code)
- Highlights "collaboration cost" angle (decreased Slack engagement)
- Limitations prevent extensive integration (solo puzzles ≠ team dev)
- Commercial bias noted (Anaconda blog by Anaconda CTO)

Technical review challenged initial 4/5 score → adjusted to 2/5.
Maintains guide rigor through minimal integration + explicit caveats.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-26 16:53:48 +01:00
Florian BRUNIAUX
1136dc683f docs: add resource-evaluations to tracked docs
- Create docs/resource-evaluations/ with 15 evaluation files
- Standardize filenames (remove date prefixes)
- Keep working docs and private audits in claudedocs/ (gitignored)
- Add resource evaluation workflow to CLAUDE.md

Files migrated:
- gsd, worktrunk, boris-cowork-video, wooldridge-productivity-stack
- remotion, nick-jensen, se-cove, self-improve-skill
- astgrep, clawdbot, prompt-repetition, uml-diagrams
- vibe-coding-rusitschka, anthropic-releases

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-26 14:02:05 +01:00