Skip to content

Get started with one governed workflow

Start with the commercial decision, not the installation command. Choose the narrowest capability that supports the work in front of you, inspect a fictional example, then let a platform owner connect governed data, provider keys, and review checkpoints only after the team agrees the output shape is useful.

Choose the first capability

01

Prepare for an account conversation

Use HCP Intelligence when the first output should be a cited public-source profile for an HCP or HCO.

HCP
02

Review one market question

Use Competitive Intelligence when the first output should be a traceable evidence run, briefing, or timeline.

CI
03

Compare coverage scenarios

Use Territory Alignment when the first output should be an assignment and route trade-off scenario.

TA
04

Prioritise the next engagement

Use Next-Best-Engagement when the first output should be a constrained plan with assigned owner and no-action reasons.

NBE
05

Equip the field from approved content

Use Field Training when the first output should be a source-grounded learning, assessment, or role-play draft.

FT
06

Prepare campaign review material

Use Campaign Studio when the first output should be a structured draft package for qualified MLR review.

CS

Match the capability to the business outcome → Inspect the fictional examples first →

Pick the install surface

Open Pharma Plugins supports two installation options. Choose the agent-harness path for the complete plugin experience, or use the published Python distribution for a Python-managed MCP server.

Option 1 · Agent harness

Install in Claude Code, Codex, or GitHub Copilot CLI

Use this option when the plugin should run inside Claude Code, Codex, or GitHub Copilot CLI. The repository installer adds the Skill and MCP tools through the harness's native plugin system.

  1. Clone the repository and enter its directory.
  2. Inspect install.sh before execution.
  3. Run bash install.sh, choose Claude Code, Codex, or GitHub Copilot CLI, then select the capability.
  4. Follow the installer verification step before using governed inputs.
git clone https://github.com/PharmaGenAI/open-pharma-plugins.git
cd open-pharma-plugins
less install.sh
bash install.sh

Open the Open Pharma Plugins repository →

Option 2 · Python distribution

Install the published package

Use this option when you want the published Python distribution or need to manage the MCP server from a Python environment. This example installs HCP Intelligence.

python -m pip install "open-pharma-plugins[hcp-intelligence]==2.4.1"

Confirm the installed entry point:

open-pharma-plugins-hcp-intelligence --version
open-pharma-plugins-hcp-intelligence --check-system

Python 3.10 to 3.13 is supported. Option 1 requires uv and uvx; install them from the official uv instructions before running bash install.sh. Keep keys in ~/.open-pharma-plugins/config, not in prompts.

Review a fictional first output before using governed data

Use the site examples as the first demo because they do not require live credentials, operational files, or provider calls.

  1. Open the capability page that matches the decision you are testing.
  2. Download the fictional input, representative output, and manifest for that capability.
  3. Confirm the output shape, review boundary, and evidence fields with the business owner before enabling real sources.

The one current runtime-generated public sample is the pinned fictional Next-Best-Engagement fixture. The other public samples are representative artifacts that show format and review boundaries, not proof that a release workflow regenerated them.

Handoff: business user and platform owner

Role First responsibility What to review before the next step
Business user Choose the workflow, define the question, and confirm the fictional output is decision-useful Business scope, accountable reviewer, and whether the output is a draft, recommendation, or evidence brief
Platform owner Install the plugin surface, configure provider keys or local directories, and constrain file access Python version, uvx availability for guided installs, key placement in ~/.open-pharma-plugins/config, and least-privilege runtime access
Qualified reviewer Validate evidence, content, privacy, policy, and approval boundaries Identity, permissions, source support, fair balance, consent, territory rules, and release suitability

Outputs and permissions to expect

Public-source research

HCP Intelligence and Competitive Intelligence

Expect structured JSON and evidence-backed summaries. Provider keys may be needed for configured web search or optional synthesis, and public APIs receive the query terms needed for retrieval.

Governed operations

Territory Alignment and Next-Best-Engagement

Expect scenario or plan outputs that depend on governed HCP, representative, consent, interaction, and assignment data. Review assignment ownership, eligibility rules, capacity assumptions, and export paths before operational use.

Approved content

Field Training and Campaign Studio

Expect drafts, scorecards, rendered assets, or review packages grounded in supplied approved sources or claims. These outputs support qualified medical, legal, and regulatory review; they are not approval evidence.

Read the trust and governance boundaries → Open the pinned technical documentation →

Utility links