Commit Graph

118 Commits

Author SHA1 Message Date
OpenCode Test
a9eaf0114f fix(external-llm): align tier defaults with industry benchmarks
- frontier: gpt-5.2 (GPT)
- mid-tier: claude-sonnet-4.5 (Claude)
- lightweight: claude-haiku-4.5 (Claude)

Prioritizes correctness over speed, aligned with MMLU/GPQA/Arena rankings.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-12 11:42:38 -08:00
OpenCode Test
f63172c4cf feat(external-llm): standardize tiers and optimize model selection
- Rename tiers: opus/sonnet/haiku → frontier/mid-tier/lightweight
- Align with industry benchmarks (MMLU, GPQA, Chatbot Arena)
- Add /external command for LLM mode control
- Fix invoke.py timeout passthrough (now 600s default)

Tier changes:
- Promote gemini-2.5-pro to frontier (benchmark-validated)
- Demote glm-4.7 to mid-tier then removed (unreliable)
- Promote gemini-2.5-flash to mid-tier

New models added:
- gpt-5-mini, gpt-5-nano (GPT family coverage)
- grok-code (Grok/X family)
- glm-4.5-air (lightweight GLM)

Removed (redundant/unreliable):
- o3 (not available)
- glm-4.7 (timeouts)
- gpt-4o, big-pickle, glm-4.5-flash (redundant)

Final: 11 models across 3 tiers, 4 model families

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-12 03:30:51 -08:00
OpenCode Test
ff111ef278 fix(external-llm): correct o3 and glm-4.7 tiers
- github-copilot/o3: opus -> sonnet-equivalent
- zai-coding-plan/glm-4.7: sonnet -> opus-equivalent

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-09 08:13:20 -08:00
OpenCode Test
bf5470ac66 fix(external-llm): correct gemini CLI model tiers
- gemini/gemini-2.5-pro: opus -> sonnet-equivalent
- gemini/gemini-2.5-flash: sonnet -> haiku-equivalent

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-09 00:02:35 -08:00
OpenCode Test
c1e3b2881d feat(external-llm): add gemini-3-flash-preview via OpenCode
Gemini 3 models via github-copilot provider (OpenCode):
- github-copilot/gemini-3-pro-preview (opus-tier)
- github-copilot/gemini-3-flash-preview (sonnet-tier)

Native Gemini CLI models unchanged:
- gemini/gemini-2.5-pro
- gemini/gemini-2.5-flash

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-08 23:52:32 -08:00
OpenCode Test
4024740b82 fix(external-llm): remove non-existent gemini-3 models
Removed gemini/gemini-3-pro and gemini/gemini-3-flash from native
gemini CLI - these models return 404 Not Found.

Remaining gemini models (via native CLI):
- gemini/gemini-2.5-pro
- gemini/gemini-2.5-flash

Note: github-copilot/gemini-3-pro-preview (via opencode) kept for now.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-08 13:49:47 -08:00
OpenCode Test
fb4cf1b035 fix(external-llm): correct opencode CLI syntax and gemini routing
- OpenCode: use `opencode run -m MODEL "prompt"` syntax
- OpenCode: set correct binary path (/home/linuxbrew/.linuxbrew/bin/opencode)
- Gemini: route long-context to gemini-2.5-pro (gemini-3 not available yet)

Tested working:
- opencode/big-pickle
- github-copilot/claude-sonnet-4.5
- zai-coding-plan/glm-4.7
- gemini/gemini-2.5-pro
- gemini/gemini-2.5-flash

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-08 13:46:38 -08:00
OpenCode Test
d2daf74fca feat(external-llm): add native gemini CLI models
Added gemini CLI models:
- gemini/gemini-3-pro (long-context, reasoning)
- gemini/gemini-3-flash (fast, general)
- gemini/gemini-2.5-pro (long-context, reasoning)
- gemini/gemini-2.5-flash (fast, general)

Updated long-context routing to use native gemini CLI (gemini-3-pro)
instead of opencode/github-copilot path.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-08 13:37:45 -08:00
OpenCode Test
e52e818686 chore: mark fc-004 external LLM integration as resolved
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-08 13:34:56 -08:00
OpenCode Test
df6cf94dae feat(external-llm): add external LLM integration (fc-004)
Implements external LLM routing via opencode CLI for:
- GitHub Copilot (gpt-5.2, claude-sonnet-4.5, claude-haiku-4.5, o3, gemini-3-pro)
- Z.AI (glm-4.7 for code generation)
- OpenCode native (big-pickle)

Components:
- mcp/llm-router/invoke.py: Main router with task-based model selection
- mcp/llm-router/delegate.py: Agent delegation helper (respects external mode)
- mcp/llm-router/toggle.py: Enable/disable external-only mode
- mcp/llm-router/providers/: CLI wrappers for opencode and gemini

Features:
- Persistent toggle via state/external-mode.json
- Task routing: reasoning -> gpt-5.2, code-gen -> glm-4.7, long-context -> gemini
- Claude tier mapping: opus -> gpt-5.2, sonnet -> claude-sonnet-4.5, haiku -> claude-haiku-4.5
- Session-start hook announces when external mode is active
- Natural language toggle support via component registry

Plan: gleaming-routing-mercury

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-08 13:34:35 -08:00
OpenCode Test
7dcb8af1bb Set default model to opus in settings.json
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 13:49:30 -08:00
OpenCode Test
6be9bf5aff Mark OpenCode transposition plan as complete
- Implementation complete (2026-01-07)
- All steps verified: backup, sync script, config, testing, docs
- 40 agents discovered (built-in + synced)
- 10 skills, 27 commands, 10 workflows synced
- Manual TUI testing pending

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 13:45:59 -08:00
OpenCode Test
7c37e9adb3 Add OpenCode sync enhancement plan and future consideration
- Add fc-047: Consider JSON minification for OpenCode instructions
- Add brainstorming plan for OpenCode Claude sync enhancements
- Add implementation status plan

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 12:15:33 -08:00
OpenCode Test
0780b4c17d Document plans index in CLAUDE.md and plans/README.md
- Add plans/ to directory structure
- Add plans/index.json to shared state files table
- Add Plans row to component formats table
- Create plans/README.md with schema and query examples

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 11:28:56 -08:00
OpenCode Test
a08dc505d9 Add plans index.json for status tracking
Central registry tracking all 17 plans with status, category, and dates.
Enables easy querying of pending vs implemented plans.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 11:27:18 -08:00
OpenCode Test
c82726b691 Add RAG JSON-to-text transformation plan
Design for improving semantic search quality by transforming JSON
structures into natural language at index time.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 11:11:34 -08:00
OpenCode Test
c14c0d843d Update ralph-loop to guardrail hooks task
Switch from completed morning-report to guardrail hooks implementation.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 11:11:27 -08:00
OpenCode Test
0fd0e74b67 Track new PA session entries
Add session records for Jan 06 and Jan 07 sessions.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 11:11:21 -08:00
OpenCode Test
1636784931 Enable tasks section in morning report, add daily archives
- Enable tasks collector in morning-report config
- Update morning.md with Jan 07 report
- Archive Jan 06 and Jan 07 reports

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 11:11:15 -08:00
OpenCode Test
c30ea2d679 Update plugin metadata from auto-updates
Plugins refreshed their cache versions and timestamps.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 11:11:08 -08:00
OpenCode Test
769391640b Add telemetry directory to gitignore
Analytics tracking data should not be version controlled.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 11:10:58 -08:00
OpenCode Test
ecf375205f Implement guardrail hooks for dangerous operation prevention
- Add PreToolUse hook intercepting Bash, Write, Edit
- Block catastrophic commands (rm -rf /, mkfs, etc.)
- Require confirmation for operations outside safe paths
- Git-aware: operations in git repos are allowed
- Session allowlist for user-confirmed operations
- Audit logging to logs/guardrail.jsonl
- Clear session allowlist on SessionEnd

Config: state/guardrails.json
Scripts: hooks/scripts/guardrail.py, guardrail-confirm.py

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 10:57:53 -08:00
OpenCode Test
f2f8a03a32 Add guardrail hooks design document
PreToolUse hooks to prevent dangerous operations:
- Intercepts Bash, Write, Edit before execution
- Contextual response (block vs confirm)
- Path-aware with git repo detection
- Session allowlist for user confirmations
- Audit logging

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 10:50:03 -08:00
OpenCode Test
630893f047 Add conditional RAG reindex after session summarization
When summarize-transcript.py extracts items to memory files, it now
triggers index_personal.py to update the RAG search index. Only runs
when items were actually added (total_added > 0) to avoid unnecessary
reindexing on trivial sessions.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-07 10:32:04 -08:00
OpenCode Test
928fa7191b Add hybrid format validation to validate-setup.sh
- Check agents/*.md, commands/*.md, workflows/*.yaml, state/*.json
- Add gtasks and other missing skills to validation list

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 13:47:46 -08:00
OpenCode Test
fae8730477 Add /tasks to README documentation
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 13:45:17 -08:00
OpenCode Test
7e563bd334 Add /tasks skill for Google Tasks
- New gtasks skill with thin wrapper around morning-report collector
- Register /tasks command with /todo, /todos aliases
- Design doc at docs/plans/2026-01-05-gtasks-skill-design.md

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 13:15:15 -08:00
OpenCode Test
9ae8ff85c3 Update local config and plugin metadata
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 13:01:02 -08:00
OpenCode Test
f9e9be62bc Add pi50 resource optimization plan, mark monitoring design complete
- New plan: Improve pi50 control plane resource usage
- Completed: Workstation monitoring design status file

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 13:00:57 -08:00
OpenCode Test
5b9a85cd37 Update state: format future-considerations, add session history
- future-considerations: Pretty-print JSON, update fc-001 to pending status
- history/index: Add recent session entries

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 13:00:52 -08:00
OpenCode Test
91733f5460 Fix gtasks OAuth scope handling and add ArgoCD docs to RAG
- gtasks: Add force_reauth option to recover from invalid_scope errors
- rag-search: Index ArgoCD documentation for semantic search

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 13:00:46 -08:00
OpenCode Test
380e2005c8 Regenerate morning report for 2026-01-05
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 13:00:40 -08:00
OpenCode Test
62050faedc Add workstation monitoring design 2026-01-05 01:31:10 -08:00
OpenCode Test
f3cb082c36 Regenerate morning report for 2026-01-04
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 23:44:40 -08:00
OpenCode Test
db0d9f97b2 Update plugin timestamps
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 23:44:33 -08:00
OpenCode Test
94603b19a5 Update session history index
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 23:44:29 -08:00
OpenCode Test
45b7e4bcf7 Improve morning report collectors and add section toggling
- Add is_section_enabled() to support per-section enable/disable in config
- Update Python path from 3.13 to 3.14 for gmail venv
- Disable tasks section by default (enabled: false in config)
- Apply code formatting improvements (black/ruff style)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 23:44:24 -08:00
OpenCode Test
7ca8caeecb Implement rag-search skill for semantic search
Add new skill for semantic search across personal state files and
external documentation using ChromaDB and sentence-transformers.

Components:
- search.py: Main search interface (--index, --top-k flags)
- index_personal.py: Index ~/.claude/state files
- index_docs.py: Index external docs (git repos)
- add_doc_source.py: Manage doc sources
- test_rag.py: Test suite (5/5 passing)

Features:
- Two indexes: personal (116 chunks) and docs (k0s: 846 chunks)
- all-MiniLM-L6-v2 embeddings (384 dimensions)
- ChromaDB persistent storage
- JSON output with ranked results and metadata

Documentation:
- Added to component-registry.json with triggers
- Added /rag command alias
- Updated skills/README.md
- Resolved fc-013 (vector database for agent memory)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 23:41:38 -08:00
OpenCode Test
c21b152de8 Add Agentic RAG design document
Design for extending Claude agent system with semantic search:
- Two indexes: personal (state files) + external docs
- ChromaDB + sentence-transformers stack
- rag-search skill with search.py CLI
- Daily systemd timer for index refresh
- Ralph loop implementation with Haiku/Sonnet delegation

Added future considerations (fc-043 to fc-046):
- Auto-sync on tool version change
- Broad doc indexing
- K8s deployment
- Query caching

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 14:08:00 -08:00
OpenCode Test
4fe8957482 Add instruction to keep repos clean by removing old branches 2026-01-04 13:29:04 -08:00
OpenCode Test
1b432f1c3f Add session summarization hook plan 2026-01-04 13:22:28 -08:00
OpenCode Test
383e2cbf38 Add morning report for 2026-01-04 2026-01-04 13:22:28 -08:00
OpenCode Test
1f5029cbb0 Update plugin timestamps and session history 2026-01-04 13:22:27 -08:00
OpenCode Test
89255cc6fa Add gitea-merge.sh for complete merge workflow
- Creates gitea-merge.sh: rebase merge, delete branch, close PR
- Updates CLAUDE.md: simplified git workflow using new script
- Adds model selection reminder to general-instructions.json
- Documents Gitea API commands in CLAUDE.md

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 13:18:23 -08:00
OpenCode Test
652ceb55f0 Add PR validation before creating PRs
- Create validate-pr.sh: runs shellcheck, JSON/YAML/Python syntax checks
- Update gitea-pr.sh: runs validation before creating PR
- Update CLAUDE.md: document PR review policy
  - ~/.claude repo: linting/validation only
  - Code repos: full code-reviewer agent review

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 13:06:31 -08:00
OpenCode Test
02f9cf7d8f Add Gitea PR automation script
- Create automation/gitea-pr.sh for auto-creating PRs via Gitea API
- Update CLAUDE.md with git workflow using the new script
- Deprecate redundant git instructions in PA general-instructions.json
- Token stored securely at ~/.config/gitea-token

Usage: gitea-pr.sh "PR Title" "Description"

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 12:52:54 -08:00
OpenCode Test
2105803594 Merge feature/user-prompt-context-injection: Add UserPromptSubmit hook for context injection 2026-01-04 12:39:59 -08:00
OpenCode Test
73400a21ab Add UserPromptSubmit hook for context injection
Injects contextual information when user submits a prompt:
- Current time with period (morning/afternoon/evening/night)
- Git branch if in a repository
- Relevant memory items based on prompt keywords (2+ matches)
- Pending decisions needing attention

Design:
- Skips short prompts (<10 chars) to not slow down commands
- 5s timeout to keep prompts responsive
- Lightweight keyword matching for memory relevance

Also updates general-instructions.json with git workflow notes.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-04 12:34:39 -08:00
OpenCode Test
56b455a074 Merge feature/session-summarization-hook: Add SessionEnd hook for automatic session summarization 2026-01-04 12:28:19 -08:00
OpenCode Test
f07022ca60 Add SessionEnd hook for automatic session summarization
Implements automatic memory extraction when Claude Code sessions end:
- Add SessionEnd hook to hooks.json with 120s timeout
- Create session-end.sh wrapper that parses hook input and runs summarizer
- Create summarize-transcript.py that:
  - Loads transcript from Claude's storage
  - Skips trivial sessions (<3 user messages)
  - Extracts paths/facts via heuristics
  - Uses Claude CLI (subscription auth) for decisions/preferences
  - Saves to memory files with deduplication
  - Updates history index with summarized flag

Uses `claude -p --model haiku` for LLM extraction, leveraging
existing subscription credentials instead of requiring API key.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-03 14:17:51 -08:00