mirror of
https://github.com/EveryInc/compound-engineering-plugin.git
synced 2026-09-28 17:12:57 +02:00
Trevin ChowandClaude Opus 4.7
b6f5e8e6e7
Convert 117 occurrences of `<category>:ce-<agent-name>` to bare `ce-<agent-name>` across 16 files in the compound-engineering plugin. The ce- prefix is sufficient for uniqueness; the `<category>:` segment was only meaningful because the Codex converter's getAgentCategory() rebuilt it from the source path, which now returns null for every agent after Unit 1's flatten. Applied via a narrow perl regex matching only the six known category names (review, research, workflow, design, document-review, docs) when followed directly by `:ce-`. Post-sweep grep for the pattern returns zero. Insertions and deletions balance exactly (117 each), confirming the rewrite was surface-only. Per docs/plans/2026-04-21-001-refactor-flatten-agents-directory-plan.md Unit 3. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ce-optimize
Run iterative optimization loops for problems where you can try multiple variants and score them with the same measurement setup.
When To Use It
Use /ce-optimize when:
- The right change is not obvious up front
- You can generate several plausible variants
- You have a repeatable measurement harness
- "Better" can be expressed as a hard metric or an LLM-as-judge evaluation
Good fits:
- Tuning memory, timeout, concurrency, or batch-size settings where you can measure crashes, latency, throughput, or error rate
- Improving clustering, ranking, search, or recommendation quality where hard metrics alone can be gamed
- Optimizing prompts where both output quality and token cost matter
Usually not a good fit:
- One-shot bug fixes with an obvious root cause
- Changes without a repeatable measurement harness
- Problems where "better" cannot be measured or judged consistently
Quick Start
- Start with
references/example-hard-spec.yamlfor objective targets - Start with
references/example-judge-spec.yamlwhen semantics matter and you need LLM-as-judge - Keep the first run serial, small, and cheap until the harness is trustworthy
- Avoid introducing new dependencies until the baseline and evaluation loop are stable
Docs
SKILL.md: full orchestration workflow and runtime rulesreferences/usage-guide.md: example prompts and practical "when/how to use this skill" guidancereferences/optimize-spec-schema.yaml: optimization spec schemareferences/experiment-log-schema.yaml: experiment log schema