diff --git a/.opencode/commands/sync-cep.md b/.opencode/commands/sync-cep.md index cb206c4..5fc69f0 100644 --- a/.opencode/commands/sync-cep.md +++ b/.opencode/commands/sync-cep.md @@ -137,6 +137,80 @@ The **only** acceptable dry-run output is the literal template above with ` + +# CHECK 2: Over-conversions (should ALWAYS be zero across ALL files) +grep -rnE '\.opencode/\.opencode/|config/opencode/\.config/|systematic\.systematic|systematic:systematic:|opencode\.ai/opencode' +``` + +If CHECK 1 returns hits on non-exception files, fix them. If CHECK 2 returns any hits, you have a double-conversion bug — investigate immediately. + ## Phase 3: Intelligent Rewrite This is the critical step that distinguishes a good import from a broken one. Every converted definition MUST go through intelligent rewrite. @@ -394,6 +474,9 @@ Before writing the file, verify: - [ ] Content matches Systematic's style (structured phases, explicit deliverables) - [ ] Agent `mode` field is set (`subagent`, `primary`, or `all`) - [ ] Agent `temperature` is appropriate for the agent's purpose +- [ ] **Batch sed fixup ran** — Phase 2c sed commands executed on all converted files +- [ ] **Grep verification passed** — Phase 2c CHECK 1 (remaining refs) and CHECK 2 (over-conversions) both clean +- [ ] **Edge cases reviewed** — Checked for mangled URLs, dead badge links, uppercase `Task()` in code blocks (see Phase 2c edge cases table) ### 3g. Discrepancy Reporting @@ -412,9 +495,6 @@ Example discrepancy report: | skills/foo/SKILL.md | 42 | Uses `~/.opencode/` path | Verify if intentional, fix to `~/.config/opencode/` if not | | commands/bar.md | 15 | Uses `AskUserQuestion` | Convert to `question` tool | ``` -- [ ] Content matches Systematic's style (structured phases, explicit deliverables) -- [ ] Agent `mode` field is set (`subagent`, `primary`, or `all`) -- [ ] Agent `temperature` is appropriate for the agent's purpose ## Phase 4: Write and Register