Use portable Agent Skills metadata and one repo-local source for
both Claude Code and OpenCode.
Deats,
- replace harness-only frontmatter and dynamic context
- make `commit-msg` runtime-aware and review-first
- add a thin OpenCode command and integration docs
- guard `np.searchsorted()` guidance before OOB indexing
- record OpenCode prompt provenance
Prompt-IO: ai/prompt-io/opencode/20260721T205228Z_58ffe487_prompt_io.md
(this patch was generated in some part by `opencode` using `gpt-5.6-sol` (`openai`))
Distilled distributed-runtime + structured-concurrency
expertise for the `tractor` actor-tree, auto-applied (not
user-invocable) when working on daemon/service arch, RPC
eps, `to_asyncio` integration, cancellation semantics or
hang/wedge/skew debugging.
Deats,
- `SKILL.md`: the core mental model incl. the post-split
actor-tree taxonomy (`datad`/`brokerd`/`emsd`/etc.),
daemon lifecycle conventions, actor-local-state hazards
and `tractor` primitive usage as deployed here.
- `gotchas.md`: symptom -> cause -> fix entries distilled
from this branch's (datad|brokerd)-split debugging (eg.
un-warmed contract caches, stale IPC resps, double/bare
log records, ib client-id collisions).
- `debug-recipes.md`: actor-system forensics incl. wedged
actor triage, hang-proof test gating and regression vs
pre-existing attribution.
(this patch was generated in some part by [`claude-code`][claude-code-gh])
[claude-code-gh]: https://github.com/anthropics/claude-code
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Gud Boi