bmad-parallel-plan

Name: bmad-parallel-plan
Author: aj-geddes

467stars73forks

How to install bmad-parallel-plan?

aj-geddes/claude-code-bmad-skills/bmad-parallel-plan

$npx -y skills add aj-geddes/claude-code-bmad-skills --skill bmad-parallel-plan

Installs into the current project.

Summary

Turns a sequential, ready-for-dev story backlog into conflict-free CONCURRENT WAVES. Builds a dependency DAG from epic order, per-story dependency maps, and Owned File/Module Scope overlaps, then topologically sorts it into parallel waves of mutually disjoint, dependency-satisfie

Files · 1

View on GitHub

SKILL.md

1	# BMAD Parallel Plan
2
3	Convert a linear backlog into waves of stories that can be developed at the same
4	time without colliding — then describe exactly how to merge them back together. This
5	skill produces a plan. Your external dev tools execute it.
6
7	Persona flavor: Winston (Architect) reasons about isolation; the workflow does the math.
8
9	## Scope (read first)
10
11	- This skill plans concurrency. It NEVER spawns worktrees, runs dev agents, writes
12	application code, runs tests, lints, or runs `git`.
13	- Inputs are planning artifacts. The only output is `parallelization-plan.md` (plus an
14	optional `dependency-graph.json` / `waves.json` for traceability).
15	- Branch names and merge order are recommendations the dev tool/human carries out.
16
17	## Inputs
18
19	\| Artifact \| Path (under output folder) \| Used for \|
20	\|----------\|---------------------------\|----------\|
21	\| Sprint status \| `sprint-status.yaml` \| story ids, epic, status, dependency lists \|
22	\| Ready stories \| `stories/{epic}.{story}.{slug}.story.md` \| Owned File/Module Scope + Dependency Maps \|
23	\| Architecture \| `architecture.md` \| semantic-conflict prevention (boundaries, shared modules) \|
24	\| Config \| `userConfig.maxParallel` (default `3`) \| wave width cap \|
25
26	Only stories at status `ready-for-dev` (or later) are eligible for a wave.
27
28	## The four steps
29
30	1. Lean on architecture for semantic safety. Read `architecture.md`. Architecture is
31	what makes parallelism safe — clean module boundaries mean two stories touching
32	different components won't create a hidden semantic conflict even if the files differ.
33	Note any shared/cross-cutting modules (auth, config, DB schema, shared types); stories
34	that touch them are high-conflict and rarely parallelizable.
35
36	2. Read each story's Owned File/Module Scope. Every ready story declares the explicit
37	list of paths it may touch. Collect `{story_id -> [paths]}`. A missing or empty scope
38	is a planning blocker — flag it; do not guess.
39
40	3. Build the dependency DAG, then topologically sort into waves. Edges come from three
41	conflict classes (see REFERENCE.md):
42	- Ordering edges — epic order (stories within an epic are usually sequential) and
43	each story's explicit Dependency Maps (`depends_on`).
44	- File-scope edges — any two stories whose Owned File/Module Scopes intersect must
45	not share a wave (an undirected conflict, resolved by lower story id first).
46	- Semantic edges — both touch a shared/cross-cutting module from step 1.
47
48	Topologically sort: wave N = all stories whose dependencies are already satisfied by
49	waves `< N` AND that are pairwise file-disjoint AND pairwise semantically safe. Cap each
50	wave at `maxParallel`; overflow rolls to the next wave (lowest id first).
51
52	4. Emit `parallelization-plan.md`. For each wave, list the ready-for-dev stories; give
53	each an isolated `git-worktree` branch name and its disjoint file scope. Then give the
54	ordered merge sequence: lowest story id first into an `integration` branch, an
55	integration review checkpoint, then a single PR `integration -> main`.
56
57	## Three intents
58
59	- Create — first wave plan from the current backlog.
60	- Update — re-plan after stories were added/finished/re-scoped (recompute the DAG;
61	exclude `done`, re-sort remaining).
62	- Validate — re-check an existing plan: confirm every wave is still file-disjoint,
63	dependency-satisfied, and within `maxParallel`; report drift.
64
65	State the intent, then proceed.
66
67	## Run the helper scripts
68
69	Both scripts are deterministic and read-only. Resolve paths via `${CLAUDE_PLUGIN_ROOT}`.
70
71	```bash
72	# 1) Build the dependency DAG (edges + conflict class) from status + story scopes
73	python3 "${CLAUDE_PLUGIN_ROOT}/skills/bmad-parallel-plan/scripts/build-dependency-graph.py" \
74	--status "<output>/sprint-status.yaml" \
75	--stories "<output>/stories" \
76	-

People who liked this also viewed

6 picks

Type

bmad-parallel-plan

467stars73forks

View on GitHub

How to install bmad-parallel-plan?

aj-geddes/claude-code-bmad-skills/bmad-parallel-plan

$npx -y skills add aj-geddes/claude-code-bmad-skills --skill bmad-parallel-plan

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# BMAD Parallel Plan
2
3	Convert a linear backlog into waves of stories that can be developed at the same
4	time without colliding — then describe exactly how to merge them back together. This
5	skill produces a plan. Your external dev tools execute it.
6
7	Persona flavor: Winston (Architect) reasons about isolation; the workflow does the math.
8
9	## Scope (read first)
10
11	- This skill plans concurrency. It NEVER spawns worktrees, runs dev agents, writes
12	application code, runs tests, lints, or runs `git`.
13	- Inputs are planning artifacts. The only output is `parallelization-plan.md` (plus an
14	optional `dependency-graph.json` / `waves.json` for traceability).
15	- Branch names and merge order are recommendations the dev tool/human carries out.
16
17	## Inputs
18
19	\| Artifact \| Path (under output folder) \| Used for \|
20	\|----------\|---------------------------\|----------\|
21	\| Sprint status \| `sprint-status.yaml` \| story ids, epic, status, dependency lists \|
22	\| Ready stories \| `stories/{epic}.{story}.{slug}.story.md` \| Owned File/Module Scope + Dependency Maps \|
23	\| Architecture \| `architecture.md` \| semantic-conflict prevention (boundaries, shared modules) \|
24	\| Config \| `userConfig.maxParallel` (default `3`) \| wave width cap \|
25
26	Only stories at status `ready-for-dev` (or later) are eligible for a wave.
27
28	## The four steps
29
30	1. Lean on architecture for semantic safety. Read `architecture.md`. Architecture is
31	what makes parallelism safe — clean module boundaries mean two stories touching
32	different components won't create a hidden semantic conflict even if the files differ.
33	Note any shared/cross-cutting modules (auth, config, DB schema, shared types); stories
34	that touch them are high-conflict and rarely parallelizable.
35
36	2. Read each story's Owned File/Module Scope. Every ready story declares the explicit
37	list of paths it may touch. Collect `{story_id -> [paths]}`. A missing or empty scope
38	is a planning blocker — flag it; do not guess.
39
40	3. Build the dependency DAG, then topologically sort into waves. Edges come from three
41	conflict classes (see REFERENCE.md):
42	- Ordering edges — epic order (stories within an epic are usually sequential) and
43	each story's explicit Dependency Maps (`depends_on`).
44	- File-scope edges — any two stories whose Owned File/Module Scopes intersect must
45	not share a wave (an undirected conflict, resolved by lower story id first).
46	- Semantic edges — both touch a shared/cross-cutting module from step 1.
47
48	Topologically sort: wave N = all stories whose dependencies are already satisfied by
49	waves `< N` AND that are pairwise file-disjoint AND pairwise semantically safe. Cap each
50	wave at `maxParallel`; overflow rolls to the next wave (lowest id first).
51
52	4. Emit `parallelization-plan.md`. For each wave, list the ready-for-dev stories; give
53	each an isolated `git-worktree` branch name and its disjoint file scope. Then give the
54	ordered merge sequence: lowest story id first into an `integration` branch, an
55	integration review checkpoint, then a single PR `integration -> main`.
56
57	## Three intents
58
59	- Create — first wave plan from the current backlog.
60	- Update — re-plan after stories were added/finished/re-scoped (recompute the DAG;
61	exclude `done`, re-sort remaining).
62	- Validate — re-check an existing plan: confirm every wave is still file-disjoint,
63	dependency-satisfied, and within `maxParallel`; report drift.
64
65	State the intent, then proceed.
66
67	## Run the helper scripts
68
69	Both scripts are deterministic and read-only. Resolve paths via `${CLAUDE_PLUGIN_ROOT}`.
70
71	```bash
72	# 1) Build the dependency DAG (edges + conflict class) from status + story scopes
73	python3 "${CLAUDE_PLUGIN_ROOT}/skills/bmad-parallel-plan/scripts/build-dependency-graph.py" \
74	--status "<output>/sprint-status.yaml" \
75	--stories "<output>/stories" \
76	-

People who liked this also viewed

6 picks

Type