You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Port the graphify skill to work with Codex and OpenClaw, not just Claude Code.
Approach
The Python backend (extract.py, build.py, cache.py, report.py) is already platform-agnostic. Only skill.md is Claude Code-specific — it uses Claude Code tool names (Bash, Agent, Read, etc.).
v3 would ship multiple skill frontends:
skill.md - Claude Code (current)
skill-codex.md - Codex tool names
skill-openclaw.md - OpenClaw tool names
Same extraction logic and Python pipeline, different orchestration layer.
Work needed
Map Claude Code tool calls in skill.md to Codex equivalents (ref: references/codex-tools.md)
Map to OpenClaw tool interface
Test each frontend against the same corpus and verify graph output matches
Update graphify install to detect platform and copy the right skill file
OpenCode support is now shipped in v0.3.0 — skill-opencode.md uses OpenCode's @mention syntax for parallel subagent extraction, and graphify install --platform opencode copies it to .config/opencode/skills/graphify/SKILL.md. Also added graphify opencode install for always-on AGENTS.md rules. Thanks for the request!
Closing — Codex, OpenCode, and OpenClaw support all shipped in v0.3.0. Each platform has its own optimized skill file and graphify install --platform <name> routes to the right config directory. Always-on hooks (graphify codex/opencode/claw install) write AGENTS.md rules for persistent graph-first behaviour.
Activity
motonari728 commentedon Apr 6, 2026
It would be great if you could also add OpenCode support.
safishamsi commentedon Apr 6, 2026
OpenCode support is now shipped in v0.3.0 —
skill-opencode.mduses OpenCode's@mentionsyntax for parallel subagent extraction, andgraphify install --platform opencodecopies it to.config/opencode/skills/graphify/SKILL.md. Also addedgraphify opencode installfor always-on AGENTS.md rules. Thanks for the request!safishamsi commentedon Apr 6, 2026
Closing — Codex, OpenCode, and OpenClaw support all shipped in v0.3.0. Each platform has its own optimized skill file and
graphify install --platform <name>routes to the right config directory. Always-on hooks (graphify codex/opencode/claw install) write AGENTS.md rules for persistent graph-first behaviour.docs: v1.0 milestone audit run #2 — passed, 31/31 in-scope satisfied,…
docs(skill): save_manifest at end of --update merge step (#545)
Prevent phantom god nodes from cross-file member-call name collisions
Prevent phantom god nodes from cross-file member-call name collisions
extract: lap-20b — bump AST_CACHE_VERSION to v2
navigate: lap-21 Graphify-Labs#2 — hint at `wu` for typed-receiver di…
peek: lap-21 Graphify-Labs#2 — curated dump for class nodes
navigate: lap-21 (Gemini Graphify-Labs#2) — `dead-ends` pivot
doc: lap-25 — brace-expand @Class.{m1,m2,m3} for multi-method docstrings
build: lap-27 Graphify-Labs#2 — vendor-path auto-tagging
fix(analyze): exclude npm dep-block keys from god-node selection (#905)
feat: zero-setup default — auto-fall back to embedded FalkorDB Lite
Fix remaining #1392 skill bugs: --directed propagation, content-only …
fix(analyze): exclude npm dep-block keys from god-node selection (Gra…
Fix remaining Graphify-Labs#1392 skill bugs: --directed propagation, …
feat(llm): opt-in backend selection, disable switch, and honest auto-…