tim-hub/powerball-harness / harness/skills/harness-work
harness-work
VerifiedExecutes plans.json tasks — solo, parallel, or breezing team mode. Use when implementing tasks or running the work loop.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 9.3 KB |
| references/breezing-mode.md | markdown | 10.7 KB |
| references/ci-failure.md | markdown | 381 B |
| references/codex-work.md | markdown | 2.9 KB |
| references/opencode-work.md | markdown | 3.0 KB |
| references/parallel-mode.md | markdown | 1.1 KB |
| references/re-ticketing.md | markdown | 1.2 KB |
| references/review-loop.md | markdown | 3.4 KB |
| references/solo-mode.md | markdown | 6.3 KB |
| references/testing-anti-patterns.md | markdown | 7.7 KB |
| references/universal-violations.md | markdown | 1.6 KB |
| references/worker-ng-rules.md | markdown | 2.4 KB |
| references/worker-self-review.md | markdown | 2.3 KB |
| templates/completion-report.md | markdown | 3.8 KB |
| templates/worker-report.v1.json | config | 1.2 KB |
references/opencode-work.md
View on GitHub# OpenCode Work — harness-work
Load this reference only when **both** conditions are true:
1. `command -v opencode` succeeds (OpenCode CLI is installed)
2. The user explicitly passes `--opencode` or asks to use OpenCode for task execution
---
## OpenCode Mode (`--opencode` explicit only)
Delegate tasks to OpenCode via the official plugin `opencode-plugin-cc` companion.
```bash
# Task delegation
bash "${CLAUDE_SKILL_DIR}/../../scripts/opencode-companion.sh" task "task content"
# Via stdin (for large prompts)
OPENCODE_PROMPT=$(mktemp /tmp/opencode-prompt-XXXXXX.md)
# Write task content to $OPENCODE_PROMPT, then:
cat "$OPENCODE_PROMPT" | bash "${CLAUDE_SKILL_DIR}/../../scripts/opencode-companion.sh" task
rm -f "$OPENCODE_PROMPT"
# Resume previous thread
bash "${CLAUDE_SKILL_DIR}/../../scripts/opencode-companion.sh" task --resume-last "continue where we left off"
```
The companion communicates with OpenCode via its HTTP REST API (`opencode serve` headless server),
providing job management, thread resume, and structured output.
Results are verified, and if quality standards are not met, fixes are applied independently.
### Combining with Other Modes
`--opencode` can be combined with other flags:
- `--opencode --breezing` → OpenCode + Breezing (OpenCode handles implementation; Lead/Reviewer structure applies)
In OpenCode + Breezing, implementation is delegated to OpenCode sessions.
OpenCode uses HTTP REST sessions rather than native `spawn_agent`/`send_input` APIs.
See `opencode-orchestration.md` (Phase 2, deferred) for parallel session management details.
---
## OpenCode Exec Review (via official plugin)
When OpenCode is available, the review loop can use OpenCode as the review path.
Record the HEAD at task start as `BASE_REF` and review the diff against that ref.
```bash
# Record base ref at task start (execute before cc:WIP update in Step 2)
BASE_REF=$(git rev-parse HEAD)
# ... after implementation completion ...
# Execute structured review via official plugin
bash "${CLAUDE_SKILL_DIR}/../../scripts/opencode-companion.sh" review --base "${BASE_REF}"
REVIEW_EXIT=$?
```
**Verdict Mapping** (opencode-plugin-cc → Harness format):
The official plugin returns structured output conforming to `review-output.schema.json`.
| opencode-plugin-cc | Harness | Verdict Impact |
|---|---|---|
| `approve` | `APPROVE` | - |
| `needs-attention` | `REQUEST_CHANGES` | - |
| `findings[].severity: critical` | `critical_issues[]` | 1 item → REQUEST_CHANGES |
| `findings[].severity: high` | `major_issues[]` | 1 item → REQUEST_CHANGES |
| `findings[].severity: medium/low` | `recommendations[]` | No impact on verdict |
AI Residuals scan runs in parallel with companion review:
```bash
AI_RESIDUALS_JSON="$(bash "${CLAUDE_SKILL_DIR}/../../scripts/review-ai-residuals.sh" --base-ref "${BASE_REF}" 2>/dev/null || echo '{"tool":"review-ai-residuals","scan_mode":"diff","base_ref":null,"files_scanned":[],"summary":{"verdict":"APPROVE","major":0,"minor":0,"recommendation":0,"total":0},"observations":[]}')"
```
The final verdict combines both results.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |