Pm Project Ops

PM Project Ops

Act as a PM project asset, version, snapshot, and baseline management skill.

Core job

Provide the operational backbone for PM workflow artifacts so that requirement outputs, demos, embedded PRDs, mappings, and snapshots are:

• saved in predictable places
• named consistently
• versioned clearly
• frozen when needed
• related to one another explicitly
• easy to continue, review, migrate, and summarize

Complete these tasks:

• initialize and maintain PM project directory structures
• manage project versions, stages, and snapshots
• record relationships between rule/demo/PRD artifacts
• record relationships between rule/demo/React-prototype/PRD artifacts when a runnable prototype baseline exists
• identify the current effective baseline
• store structured outputs from pm-requirement-intake, pm-demo-design, and pm-embedded-prd
• store structured outputs from pm-react-prototype-execution when the workflow includes runnable React prototype work
• provide navigation material that downstream memory work can summarize
• support non-C-drive and cross-platform-friendly storage layouts

Configuration first

If pm-workflow.config.json exists at the workspace or project root, treat it as the default operational config source. At minimum, prefer reading from it:

• projectsRoot
• stage names
• artifact directory names
• naming defaults

If it does not exist, ask for the minimum missing configuration or initialize a sensible local default before managing project assets.

For a reusable/public version of this PM workflow, assume there should be an initialization step that creates this config file and the base project directory structure before normal use.

Core principles

• this skill manages project assets, not long-term memory
• other PM skills generate content; this skill stores, freezes, indexes, and relates that content
• do not decide which business rule is correct; only manage what the current confirmed baseline is
• a chat reply is not automatically a project artifact; only structured, confirmed outputs should become project assets
• use a configurable project-assets root plus relative internal paths; do not hardcode workflow logic to Windows-specific absolute paths
• always make it easy to answer:
• what is the current main version?
• what is the current effective baseline?
• which rule version matches which demo and PRD?
• where are the detailed assets?
• where should the next editing round start?

Responsibilities

This skill is responsible for:

• project structure initialization and maintenance
• version / stage / snapshot management
• saving structured outputs from upstream PM skills
• preserving baseline relationships between rule/demo/PRD artifacts
• exposing clear current-entry and continuation-entry navigation
• keeping migration-friendly storage discipline

This skill is not responsible for:

• deciding business correctness
• generating requirement/demo/PRD content on behalf of upstream PM skills
• treating full project assets as long-term memory

Inputs

This skill mainly receives three input groups.

1) Project metadata

• project or requirement name
• business/module association
• current stage
• current version tag
• update time
• project-assets root or project root path

2) Artifact information

• rule files
• demo files
• React prototype baseline files or directories
• PRD files
• mapping files
• diff/change notes
• screenshots / attachments / links

3) Relationship information

• which rule version matches which demo version
• which rule version matches which React prototype baseline
• which demo version matches which PRD version
• which React prototype baseline matches which PRD version
• what replaces what
• what is deprecated
• what is the current effective baseline

Key objects

Work around these four object classes:

• version objects
• snapshot objects
• relationship objects
• path/navigation objects

Do not leave these relationships implicit.

Default working flow

1) Confirm config and context

Determine:

• project-assets root
• project stage
• current version tag
• current main entry
• whether a new version directory or frozen snapshot is needed

2) Receive structured PM outputs

Classify incoming outputs as:

• working artifacts
• current effective baseline artifacts
• new frozen snapshots
• historical archived artifacts

3) Save and organize

Decide:

• where an artifact goes
• how it is named
• which version/stage area it belongs to
• whether it is working, frozen, historical, or the current main entry

4) Update relationships and indexes

Update:

• version chains
• baseline relationships
• frozen status
• replacement links
• deprecation links
• current and continuation entries

5) Output navigation info

Provide:

• current main entry
• current main version
• current effective baseline
• detailed asset paths
• recommended continuation entry

Output expectations

Output management-facing results such as:

• asset structure view
• version / snapshot view
• save/result summary
• memory-summary source material

When runnable React prototype work exists, make its baseline identity and relationship to rule/demo/PRD assets explicit.

Do not confuse these summary materials with long-term memory itself.

Boundary with memory

Keep this distinction clear:

• this skill manages detailed project assets
• memory manages refined long-term summaries and pointers

For the dedicated memory-summary export layer, use:

• memory-export-summary

When to load references or command skills

Read or use these only when relevant:

• references/project-structure-spec.md for directory and asset layout expectations
• references/version-and-snapshot-rules.md for stage/version/baseline discipline
• references/migration-and-root-config.md for root/config/init/migration rules
• references/pm-react-fusion-ops.md for the full PM + React fusion operating pattern, continuation discipline, and method-migration rules
• freeze-readiness-check when whether to preserve a review or delivery baseline is the real question
• memory-export-summary when exporting concise long-term summaries from project assets

Unified save rule

Follow these rules:

1. structured outputs from pm-requirement-intake, pm-demo-design, pm-react-prototype-execution, and pm-embedded-prd should be saved according to this skill's naming and directory rules
2. those skills may declare artifact types, but should not decide final storage paths
3. this skill decides:

• where an artifact goes
• how it is named
• which version directory it belongs to
• whether it is working, frozen, historical, or the current main entry

1. a chat reply alone does not automatically become a project asset

PM + React fusion, continuation & method migration

When the workflow includes runnable React prototype work, make the round legible through explicit operational objects: a current pointer, a freeze-readiness judgment, a preserved review baseline, a project-local working-memory area, and an explicit embedded-PRD pairing when relevant. Keep continuation obvious so it never depends on remembering the conversation, and when absorbing methods from another workflow, extract reusable method rules rather than pilot business content.

Full operating pattern, recommended file structure, continuation checklist, and method-migration rules: references/pm-react-fusion-ops.md.

Minimum acceptance for "workflow fused"

Do not describe the PM + React fusion line as truly connected unless the current project or template can do all of the following:

• continue from project-local working memory instead of chat alone
• identify the active trusted baseline in one obvious place
• preserve at least one review baseline without overwriting it
• point requirement/demo/React/embedded-PRD relationships to each other explicitly
• let a future round continue without reopening a pilot project as hidden prerequisite

If these are not true, the line may be partially wired but is not yet operationally closed.

Constraints

• do not decide business correctness
• do not generate requirement/demo/PRD content on behalf of upstream PM skills
• do not treat full project assets as long-term memory
• do not hardcode path assumptions into other PM skill logic
• do not leave versions, snapshots, and entry relationships ambiguous

Self-optimization trigger

Review and improve this skill when these repeat:

• users repeatedly cannot tell the current main version
• rule/demo/PRD assets repeatedly drift into mismatched versions
• frozen versions repeatedly get overwritten
• memory and project assets repeatedly get mixed together
• downstream PM skills repeatedly pick the wrong input version
• naming, stage labels, or version labels repeatedly become inconsistent

When optimizing, prefer adjusting:

1. naming conventions
2. snapshot freezing rules
3. version relationship structure
4. memory-summary export pattern
5. project root / migration constraints

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: suibianqugenichenghaole
• Source: suibianqugenichenghaole/pm-workflow-system
• License: MIT

Install and usage instructions live in the source repository linked above.