← Back to Plugins
Integration

Biotech Corpdev Agents

ronnyhamptontech-hub By ronnyhamptontech-hub 👁 8 views ▲ 0 votes

14 Openclaw plugins for biotech corp-dev intelligence: SEC filings, clinical trials, FDA/EMA, IP, literature, BD deals, capital markets, KOLs, valuation, and diligence memos. Public-API-only data sources.

GitHub

README

# Biotech Corp Dev Agent Kit

A Claude Code subagent system for continuous search and evaluation of biotech companies, assets, and deals. Designed for corporate development teams who need to monitor public filings, clinical trials, regulatory actions, IP, literature, deals, and capital markets — and turn the resulting signal into diligence memos and daily briefs.

## What's in here

- **`.claude/agents/`** — 14 specialized subagents (system prompts) that Claude Code can delegate to
- **`.claude/commands/`** — slash commands that orchestrate multi-agent workflows
- **`CLAUDE.md`** — shared conventions every agent reads (watchlist paths, output schema, provenance standard, severity tagging)
- **`config/watchlists/`** — your editable lists of companies, therapy areas, targets, and KOLs (the "first-class objects" that drive every agent)
- **`state/`** — per-agent state for differential alerts (what's been seen before, so agents only surface what changed)
- **`briefs/`** — dated output directory; every agent run writes here

## Install

1. Copy this entire folder into the root of a Claude Code project (or merge `.claude/` into an existing one).
2. Edit `config/watchlists/*.yaml` with your actual companies, therapy areas, targets, and KOLs.
3. Edit `config/preferences.yaml` for alert thresholds and verbosity.
4. From inside the project, run `claude` and try `/morning-brief` to test orchestration.

Subagents are auto-discovered. To list them: `claude agents`. To invoke explicitly: `Use the sec-filings-sentinel subagent to check today's filings.`

## The 14 agents

**Tier 1 — Continuous monitoring (run daily, diff against state):**
1. `sec-filings-sentinel` — SEC EDGAR + foreign equivalents
2. `clinical-trials-watcher` — ClinicalTrials.gov + EU CTR + WHO ICTRP
3. `regulatory-intelligence` — FDA/EMA/PMDA/MHRA actions and calendars
4. `patent-ip-monitor` — USPTO/EPO/PTAB filings, grants, litigation
5. `scientific-literature-tracker` — PubMed, preprints, conference abstracts

**Tier 2 — Deal & market intelligence:**
6. `bd-licensing-tracker` — BD/licensing deal terms and comp tables
7. `capital-markets-agent` — biotech IPOs, follow-ons, PIPEs, runway
8. `private-company-discovery` — stealth and emerging private companies

**Tier 3 — Deep diligence & analysis:**
9. `financial-analysis` — full financial models from filings
10. `pipeline-valuation` — risk-adjusted NPV (rNPV) and SOTP
11. `competitive-landscape` — target/MoA/indication competitive maps
12. `kol-network` — KOL identification, networks, sentiment

**Tier 4 — Synthesis:**
13. `daily-brief-orchestrator` — pulls from all agents into one prioritized brief
14. `diligence-memo-builder` — on-demand structured memo from agent outputs

## How the agents work together

```
                                    ┌─────────────────────┐
                                    │   Watchlists +      │
                                    │   Preferences       │
                                    └──────────┬──────────┘
                                               │  (read by all)
       ┌───────────────────┬───────────────────┼───────────────────┬───────────────────┐
       ▼                   ▼                   ▼                   ▼                   ▼
   Tier 1              Tier 1              Tier 1              Tier 2              Tier 3
   (monitors)          (monitors)          (monitors)          (deals)             (analysis)
       │                   │                   │                   │                   │
       └───────────────────┴───────────────────┼───────────────────┴───────────────────┘
                                               │
                                               ▼
                                  ┌────────────────────────┐
                                  │ daily-brief-orchestrator │
                                  └────────────┬───────────┘
                                               │
                                               ▼
                                  ┌────────────────────────┐
                                  │   diligence-memo-builder │  (on-demand)
                                  └────────────────────────┘
```

## Design principles (encoded in `CLAUDE.md`)

- **Watchlists as first-class objects.** Edit once in `config/`, every agent reads them.
- **Provenance on every claim.** Source URL + retrieval timestamp + snippet, or it doesn't ship.
- **Differential alerts.** After backfill, agents only surface what changed (state diffs).
- **Human-in-the-loop triage.** Agents propose; you dispose. Every brief includes a feedback section.
- **No silent inference.** If a number is calculated, the formula and inputs are shown.
- **Conference mode.** During ASCO/JPM/ASH/AACR, agents shift to higher-frequency tagged output.

## Caveats worth knowing

- **Public-data versions of agents 6 (BD deals) and 8 (private discovery) are weaker than licensed versions.** If you have access to PitchBook, Cortellis, BioPharma Deals, or Evaluate, swap the data source sections of those two agents to point at their APIs.
- **API rate limits matter.** PubMed wants an API key (and email), EDGAR has a 10 req/sec limit, FDA OpenAPI has its own limits. The agents will write helper scripts in `scripts/` as needed.
- **Some sources require login** (Cortellis, BioPharma Deals, EvaluatePharma). Those agents will tell you when they need a credentialed fetch and let you paste content.
- **These agents do not give investment advice.** They surface, structure, and analyze — you decide.
integration

Comments

Sign in to leave a comment

Loading comments...