B — Some risk signals Graded October 16, 2026Minor concerns worth a quick check.Based on public adoption and transparency signals as of October 16, 2026. Our opinion, not a code audit or security guarantee.
AI agent skill · General
pm-skills
PM Skills Marketplace: 100+ agentic skills, commands, and plugins — from discovery to strategy, execution, launch, and growth.
Bottom line: Signal-based grade (4/7 points): very widely starred on GitHub, public source code. This reflects adoption and transparency, not a code audit.
- Popularity score
- 49/100
- Rank
- #131 of 500
- GitHub stars
- 27.1k
What it does
PM Skills Marketplace provides 100+ pre-built agentic skills, commands, and plugins designed to automate product management workflows across discovery, strategy, execution, launch, and growth phases. It enables AI agents to execute PM-specific tasks programmatically, from market research to product launches.
Who it’s for
Product managers, AI builders, and teams looking to automate PM workflows with agentic capabilities. Best suited for organizations comfortable evaluating and integrating community-contributed skills into their AI agent infrastructure.
Copy-paste setup
Check these steps against the project’s own instructions before running anything.
# Verify against the repo README for latest installation method
# Likely installation via npm:
npm install pm-skills
# Or for Claude Desktop integration, check README for mcpServers config
# Example structure (verify in repo):
# Add to ~/.config/Claude/claude_desktop_config.json
Please verify the exact setup steps and any required environment variables against https://github.com/phuryn/pm-skills (opens external site) README.
Risks
- Credential scope: Skills may require API keys or authentication tokens (e.g., for market research tools, analytics platforms); improper scoping could expose sensitive integrations.
- Prompt injection: Community-contributed skills may not be hardened against adversarial inputs that redirect agentic execution.
- Supply-chain trust: 100+ skills from a single repository creates a large attack surface; individual skills may lack independent security review.
- Data exfiltration: Skills accessing customer/product data (discovery, strategy) could inadvertently log or transmit sensitive information if not audited.
- Maintenance gaps: Community contributions may become unmaintained; outdated skills could harbor vulnerabilities or break with API changes.
How to stay secure
Least-privilege permissions
- Scope API keys and tokens to individual skills with minimal required permissions; never share a master credential across multiple skills.
- Use read-only credentials for discovery, market research, and analytics skills; restrict write access only to skills that must execute actions (launch, deployment).
- Implement API key rotation every 90 days and revoke credentials immediately when a skill version is deprecated or removed from your environment.
- Require explicit allow-lists for external services each skill can call; deny outbound requests to unlisted domains by default.
Network isolation
- Run pm-skills in a sandboxed environment with egress filtering; allow only documented API endpoints (e.g., market research, analytics, messaging services).
- Segregate skill execution from production databases and customer data stores; use read-only replicas or anonymized datasets for testing and discovery workflows.
- Deploy a Web Application Firewall (WAF) or API gateway in front of pm-skills to detect and block malformed or suspicious skill requests before execution.
Monitoring
- Log all skill invocations with input parameters, API calls, and outputs; flag skills that make unexpected external requests or access unscoped resources.
- Alert on high-cardinality API calls, unusually large data transfers, or repeated failures that may indicate probing or exfiltration attempts.
- Audit community-contributed skills quarterly for abandoned maintenance, outdated dependencies, or new CVEs; remove or isolate skills without active security updates.
Red flags to watch for
- Skills requesting broad scopes (e.g., "all analytics APIs") or overly generic credentials; demand justification and narrow scope.
- Community contributions with no maintainer, no changelog, or "last updated >12 months ago"—treat as unmaintained and disable.
- Skill behavior that deviates from documented purpose: e.g., discovery skills making write calls, or strategy skills transmitting raw customer data to external logging services.
- Input validation failures where skills echo or execute user-supplied prompts without sanitization; test each new skill with injection payloads before production use.
Alternatives
- Langchain Agent Toolkit (opens external site) — General-purpose agent skills with broader ecosystem support
- OpenAI GPT Store / Actions (opens external site) — Pre-built integrations curated by OpenAI with varying PM relevance
- Zapier AI Actions (opens external site) — Workflow automation with AI, broader use-case coverage but less PM-focused
- In-house skill development — Custom agentic skills built and maintained internally for tighter security and compliance control