songoao25/dsh-virtual-product-team
Product Team Mode - a DeepSeek Harness agent preset: user-led conversation with a virtual product team (PM to Engineer to QA to Release) walking you from idea to shipped product
Listed
4
Skill
Bundle verified
What it does
Virtual product team agent preset: user-led conversation walks you through the full 12-step pipeline from idea to shipped, promoted and operated product, with staged gates and per-stage skills.
Best for
- Nontechnical founders or solo builders who want guided decisions from idea validation through release and operations.
- Product work that benefits from approval gates, documented outputs, and distinct PM, engineering, QA, and release stages.
- Teams building DSH modes or plugins that want the preset's Cordis authoring skills and explicit DSH-specific tooling.
Not ideal for
- Small, isolated tasks that do not need the full product lifecycle or repeated stage approvals.
- Users who require a visual progress panel, since the documented version is conversation-only.
- Existing nonblank conversations where switching into the preset is required, because DSH only permits mode changes in a blank conversation.
README
Product Team Mode
| English | 中文 |
Turn DeepSeek Harness into your virtual product development team. Say “I have an idea” and the AI walks you through the full pipeline — Product Manager → Engineer → QA → Release Engineer — from idea to shippable product. You only talk and make decisions. No technical knowledge required.
What it is
Product Team Mode is a conversation mode (agent preset) for DeepSeek Harness (DSH). Inside this mode:
- You say: “I have an idea, I want to build XX”
- The AI automatically starts the pipeline: it interviews you like a product manager to clarify the idea → writes a requirements document for your review → designs the technical plan → implements it → runs QA and security audit → prepares release materials
- Each stage finishes with a report to you; you approve before it moves to the next stage (stage-gate control)
From start to finish, you never write code, never learn the process, and never need to remember technical jargon.
The eight stages (all 12 phases covered)
| Stage | What happens | Output |
|---|---|---|
| 1. Idea validation | Research market / competitors / feasibility | Validation conclusion |
| 2. Product definition & requirements | Positioning + concrete requirements with acceptance criteria | Product definition + PRD |
| 3. Technical design | Technical plan and task breakdown | Tech design + task list |
| 4. Development & quality | Implement + test + security audit | Code + audit report |
| 5. Release & deploy | Prepare distributable artifacts (GitHub standards) + go live | README / version / Release / deploy check |
| 6. Promotion & cold start | Launch kit (video script / article / channels) | Promo materials |
| 7. Operations & growth | Metrics dashboard, feedback channels, growth actions | Operations plan |
| 8. Iteration & maintenance | Feedback pool, roadmap, then loop to the next round | Roadmap |
Installation
Prerequisites: DeepSeek Harness installed (dsh available in PATH).
git clone https://github.com/songoao25/dsh-virtual-product-team.git
cd dsh-virtual-product-team
./install.sh
Then start a new conversation, pick Product Team Mode in the mode picker, and simply say: “I have an idea…”.
Note: DSH only allows switching modes in a blank conversation, so start a new one first.
Uninstall
cd dsh-virtual-product-team
./uninstall.sh
After uninstalling, new conversations return to the default mode. No leftovers, other modes are untouched.
Install as a skill pack (optional)
Since v1.2.0 the repo also declares a DSH bundle, so you can install it as a skill pack on a profile:
dsh plugin --profile <profile-name> add dsh-virtual-product-team
This gives every session of that profile on-demand access to the 8 pipeline skills (idea validation → … → iteration), with no preset.
There are two ways to get the skills — pick whichever fits:
| Method | What you get | Command |
|---|---|---|
| Full mode (recommended) | Preset + persona + tools + the 10 skills (8 stage + 2 authoring) | ./install.sh |
| Skill pack only | Just the 8 stage skills, available to any session of a profile | dsh plugin --profile <p> add dsh-virtual-product-team |
The two share the same skill content, so you only need one. Full mode is recommended: one install covers everything the mode offers. (The bundle installs only the skill layer — the preset, persona, and tools still come from ./install.sh.)
FAQ
Q: Is it a plugin? A: It is primarily a conversation mode (preset). Since v1.2.0 it is also shipped as a DSH bundle so the 8 pipeline skills can be installed as a skill pack per profile; this is complementary, the full mode still installs with ./install.sh.
Q: Do I need to know tech? A: No. The AI makes all technical decisions. You just answer questions and approve.
Q: Does it affect my existing conversations/modes? A: No. It’s just one additional mode. Standard, Creator, and other modes remain unchanged.
Q: What are the v1 limitations? A: Conversation-only, no visual progress panel. For GitHub publishing, the release stage asks you which local AI assistant to use and then handles the commit/tag/Release with it.
Q: Can it develop DSH modes/plugins (like Creator mode)? A: Yes, since v1.1.0. The mode ships the same self-modification toolset and the two official Cordis authoring skills as Creator mode, so your product team can build new DSH modes and plugins too. That toolset carries shell-level trust and is only used when you explicitly ask for a DSH-specific product.
License
MIT © songoao25
Frequently Asked QuestionsFAQ
Use the verified command dsh plugin --profile default add github:songoao25/dsh-virtual-product-team in a DSH-enabled shell. The command resolves the public package metadata and keeps the plugin attached to the catalog identity shown on this page.
Compatibility follows the bundle and profile status shown above. If a profile is not detected, keep the plugin disabled there and check the repository documentation before enabling it in production.
The GitHub link and activity metadata are the source of truth for releases and maintenance. Revisit this page after a new release to confirm the catalog has observed the latest version.