Skip to content

Adonis0123/adonis-skills / skills/agent-dispatch

agent-dispatch

Flagged

Choose the lightest agent setup for a coding, review, drafting or design-research task: do it yourself, fan out to 2-3 workers with independent parts across the Claude Code, Codex, Grok Build, Cursor or other CLIs and accounts on this machine, run a scoped architecture loop, or send one brief to independent researchers; pick a cross-model reviewer, test budget and cleanup; track dispatched work, external waits and user decisions on a Run Board. Use when work may need delegation, or the user asks to dispatch, fan out, parallelize, run several agents or accounts, get another AI's review or independent design opinions, or asks how dispatched work is going. Chinese triggers: 派活、并行、要不要多开几个 agent、分给多个 Claude 账号 / codex / grok、找另一个 AI 审、对比优秀设计、多方意见、进度看板、派出的任务进行到哪了。Not for unrelated read-only questions, trivial edits, research with no delegation, a single review of an existing diff (agentic-review-handoff), or the machine-wide session list (sideby sessions). Inside another named workflow, only size its workers.

FileKindSize
GLOSSARY.mdmarkdown2.1 KB
SKILL.mdskill md27.8 KB
evals/README.mdmarkdown7.6 KB
evals/README.zh-CN.mdmarkdown7.5 KB
evals/bench/compare.pyscript5.3 KB
evals/bench/run_bench.pyscript12.3 KB
evals/bench/tasks.jsonconfig6.6 KB
evals/bench/test_bench.pyscript4.4 KB
evals/build_plugin_eval.pyscript11.3 KB
evals/evals.jsonconfig31.4 KB
evals/results.mdmarkdown15.8 KB
evals/test_eval_harness.pyscript3.5 KB
evals/trigger-eval-set.jsonconfig1.6 KB
fixtures/probe-full.jsonconfig1.6 KB
fixtures/probe-single.jsonconfig690 B
fixtures/run-midrun/B9.jsonlother191 B
fixtures/run-midrun/deploy-log.txtmarkdown42 B
fixtures/run-midrun/run.jsonconfig1.8 KB
fixtures/run-sessions/B9.jsonlother191 B
fixtures/run-sessions/deploy-log.txtmarkdown42 B
fixtures/run-sessions/ended.jsonlother95 B
fixtures/run-sessions/run.jsonconfig2.1 KB
fixtures/run-sessions/sessions.jsonconfig804 B
fixtures/session-index.jsonconfig401 B
fixtures/sideby-stub/sessions.jsonconfig1.5 KB
fixtures/sideby-stub/sidebyexecutablescript4.1 KB
references/accounts.mdmarkdown10.4 KB
references/context-budget.mdmarkdown2.9 KB
references/fallbacks.mdmarkdown4.7 KB
references/launch.mdmarkdown16.1 KB
references/measure.mdmarkdown3.3 KB
references/research-brief.mdmarkdown1.3 KB
references/run-board.mdmarkdown48.7 KB
references/task-spec.mdmarkdown5.0 KB
scripts/board.cssother12.6 KB
scripts/board_interaction.jsscript3.7 KB
scripts/board_markdown.jsscript2.9 KB
scripts/board_render.pyscript17.2 KB
scripts/dispatch_board.pyscript95.3 KB
scripts/isolated_case.pyscript1.6 KB
scripts/probe_agents.pyscript12.9 KB
scripts/sideby_cli.pyscript3.2 KB
scripts/sideby_sessions.pyscript3.3 KB
scripts/test_billing_and_quota.pyscript10.9 KB
scripts/test_board_interaction.jsscript2.1 KB
scripts/test_board_markdown.jsscript2.9 KB
scripts/test_board_reload_state.jsscript2.7 KB
scripts/test_bug_regressions.pyscript4.7 KB
scripts/test_dispatch_board.pyscript68.9 KB
scripts/test_harness_limits.pyscript11.7 KB
scripts/test_lane_states.pyscript7.5 KB
scripts/test_probe_agents.pyscript19.2 KB
scripts/test_review_fixes.pyscript6.7 KB
scripts/test_rounds_and_watch.pyscript7.3 KB

references/task-spec.md

View on GitHub
# Worker task spec

Copy this template into one file per worker. Every field is required; write `none` rather than deleting a field.

```markdown
# <task title>

## Target

Repository, worktree path, the lane's short branch and base (shared base, or the branch of the lane it depends on), and the files or component in scope. When the target is not in a repository, or is a live file other sessions read (global rules, host config), give no worktree: the Worker writes a proposed copy under the Run or report directory, and the Coordinator swaps it in, verifies it and keeps a backup to restore.

## Change

The concrete result to produce, in observable terms.

## Original request

The user's own words for this work, quoted, or the path of the requirement file. The Coordinator's wording above is a retelling; when it disagrees with the quote, the quote wins and the Worker reports the gap under `Deviations:`.

## Ruled out

Approaches already tried or rejected, with the reason, and facts already checked, so the Worker does not redo them. `none` when there are none.

## Constraints

Invariants, compatibility rules, repository instructions to read first (nearest AGENTS.md / CLAUDE.md), and environment entry points such as dev URLs.

## Ownership

- May edit: <exact files or globs> (register the same list with `lane add --owns`)
- Must not edit: <files other workers own, shared packages, generated files>
- Pick one, delete the other: "No commit" (the default) or "Local commits on <short-branch> only, hooks must pass" (only when the user authorized commits for this Run; a worktree alone is not that authorization).
- No push, merge into other branches, deploy, publish, or messages to other sessions.
- Scope does not widen: when the work needs a file outside May edit, or a review asks for a change beyond the Objective, stop and return `OUTCOME: handback` with the reason; the Coordinator starts a new Lane or rewrites this spec.

## Acceptance

The tests, commands, or page evidence that prove completion. Run only affected tests; no full suite unless listed here. Examples in the request are a floor, not the spec: also test boundaries the examples skip (empty, zero, negative, NaN, rounding of halves, binary floating point).

## Progress

After each finished step, overwrite <run dir>/<lane>-progress.md with: steps done (with the evidence each produced), the step in progress, steps left, and decisions so far. Keep it under 30 lines. When you hit a quota or rate limit, or the Coordinator asks you to hand off, finish or abandon the current step, update this file and return `OUTCOME: handback`. A fresh session continues from this file and the spec, not from the start.

## Return

Write the full report to <run dir>/<lane>-report.md and a summary of at most 10 lines to <run dir>/<lane>-summary.md, in the user's language (Chinese for a Chinese-speaking user). The summary has two parts:

1. **First sentence, for the user**: what now works or what was found, in plain words for a reader who does not code (e.g. "导出按钮改成了 Export video,中英文都已更新。"). End it with a full stop (。 or .) on its own line: the Board card shows only this sentence and cuts at the first full stop, so no `OUTCOME:`, ids or file paths in it.
2. **Remaining lines, for the Coordinator**, one item per line, each written even when empty (`none`): `Decisions:` choices you made that the spec did not settle; `Deviations:` anything you did differently from the spec, and why, including any Constraint you could not keep (check each Constraint before you write this line); `UNVERIFIED:` checks you could not run or that failed; `Question:` the one concrete question a person must answer before this work can go on, with your recommended answer (the Coordinator records it unchanged with `ask add --lane <lane>`; do not leave it only in the full report); `Files:` how many files you touched and any outside May edit. The Coordinator integrates from these lines without reading the full report, so a decision missing here is lost.

Your final message is that summary plus both paths, nothing else.

The full report has:

1. One machine line on its own: `OUTCOME: succeeded`, `OUTCOME: failed`, `OUTCOME: blocked` (waits on the `Question:` in the summary) or `OUTCOME: handback` (the work is beyond this spec's scope; say why).
2. Touched-file list and `git diff --stat`.
3. Evidence: each command run and its result.
4. Open questions or decisions you made, one line each.
5. Confirmation that no test runner, type checker, or server you started is still running.
```

Keep the spec self-contained: the worker cannot see the coordinator's conversation. For a follow-up round, resume the writer's session and send only the delta (review findings, the coordinator's decisions, what to change) plus the original spec path.

Register both files with `lane set --summary-file <lane>-summary.md --result-file <lane>-report.md`. The Coordinator reads the whole summary (both parts), not the full report; the full report goes to the Board's report page, or to a subagent when a detail must be checked (see `context-budget.md`).
finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the workStars: 322kUpdated Claude CodeCodexVerifiedInstall
systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixesStars: 322kUpdated Claude CodeCodexVerifiedInstall
test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation codeStars: 322kUpdated Claude CodeCodexVerifiedInstall
skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.Stars: 197kUpdated Claude CodeCodexVerifiedInstall