OpenCode
Confabulous has first-class support for OpenCode sessions.
What gets parsed
Section titled “What gets parsed”- Full conversation history (user, assistant, tool calls, tool results).
- Per-message token counts (input, output, cache read, cache write, reasoning).
- Model identifier per message — OpenCode can run models from any of its supported providers.
- Tool invocations and their arguments.
- File edits and reads.
Analytics cards
Section titled “Analytics cards”Each OpenCode session produces these cards:
- Tokens — input/output/cache breakdown.
- Cost — based on the model’s published pricing.
- Tools — which tools were called and how often.
- Conversation — turn structure, active time, message counts.
- Repo activity — files touched, language breakdown.
Subagent aggregation
Section titled “Subagent aggregation”When an OpenCode session spawns subagents (explore, build, etc.), Confabulous aggregates the main thread plus every subagent thread for most analytics cards. The Conversation card stays main-only by design (it reflects user-perceived turn structure, not subagent reasoning). Subagent uploads require a recent CLI and backend; older deployments simply skip them.
Pricing
Section titled “Pricing”Because OpenCode runs models from many providers, each session is priced against the published rates for whichever model produced it.
Other supported providers
Section titled “Other supported providers”Confabulous treats every provider as a first-class citizen. Claude Code, Codex, and Cursor are also supported today. New providers slot into the same sync, storage, and analytics pipeline.