flux-parallel-dispatch
/flux-parallel-dispatch
Use when 2 or more investigations, scaffolds, reviews, or fixes are independent enough to run in parallel without shared files, shared state, or sequential dependency.
- Author
- nairon-ai ↗
- Repository
- flux ↗
- Licence
- MIT
- Last repository change
- 12.05.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 12.05.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Use when 2 or more investigations, scaffolds, reviews, or fixes are independent enough to run in parallel without shared files, shared state, or sequential dependency.
Original author description
Use when 2 or more investigations, scaffolds, reviews, or fixes are independent enough to run in parallel without shared files, shared state, or sequential dependency.
How it works
- When to Use
- Decision Test
- Dispatch Pattern
- Prompt Requirements
- Flux Fit
- Gotchas
SKILL.md preview
--- name: flux-parallel-dispatch description: Use when 2 or more investigations, scaffolds, reviews, or fixes are independent enough to run in parallel without shared files, shared state, or sequential dependency. user-invocable: false --- # Parallel Dispatch Dispatch parallel subagents only when the work cleanly separates. Speed comes from independence, not from spawning aggressively. ## When to Use Use this skill when: - Multiple failing areas have different likely root causes - `flux:prime` or `flux:scope --explore` needs several independent scouts - Review work can be split by subsystem or concern - A large task has disjoint write sets and one agent can integrate later Do not use this skill when: - The next local step is blocked on a single urgent result - Agents would edit the same files or depend on the same evolving state - Failures may share one root cause ## Decision Test Parallelize only if all of these are true: 1. Each unit of work can be described in one paragraph 2. Each agent can succeed with task-local context only 3. The write sets are disjoint, or the work is read-only 4. One result is unlikely to invalidate the others If any check fails, keep the work local or run sequentially. ## Dispatch Pattern 1. Split by problem domain, not by arbitrary count 2. Keep one owner in the main thread for integration 3. Give each agent explicit scope, constraints, and expected output 4. Ask for summaries, not long transcripts 5. Verify and integrate before claiming success Good task shapes: - One failing test file - One scout domain - One review concern - One bounded implementation slice with clear file ownership Bad task shapes: - "Fix all the tests" - "Review everything" - "Refactor the auth system and update related code" ## Prompt Requirements Each dispatched task should include: - Goal - Exact scope - Files or subsystem owned - Constraints - Expected return format Example: ```text Investigate auth middleware failures in tests/auth-middleware.test.ts only. Find root cause, fix that file or its direct implementation dependencies, and report: - root cause - files changed - verification run Do not edit billing or session code. ``` ## Flux Fit Best fits: - `flux:prime` scouts - `flux:scope --explore` - multi-domain debugging during `flux:work` - independent fix lanes inside review workflows ## Gotchas - Parallel is wrong when one failing domain may be a symptom of another - Do not dispatch urgent critical-path work that you need immediately - Do not let multiple agents edit the same hot files without explicit ownership - Always re-run shared verification after integration
Versions
The database has no other skills named flux-parallel-dispatch.
Database snapshot: 13.11.2026 · package version bf18698754d6 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 96dispatching-parallel-agents
Use for 2+ independent tasks with no shared state or sequential dependency. - 100dispatch-multiple-agents
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies. Dispatch subagents to work concurrently. - 90dispatching-parallel-agents
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies - 92dispatching-parallel-agents
当同时面对 2 个及以上相互独立、无共享状态且无顺序依赖的任务时必须使用本技能 - 100dispatching-parallel-agents
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies - 100dispatching-parallel-agents
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies