AgentPMSAFE
MCP server for the planning and execution of AI-assisted development projects.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
[](https://smithery.ai/server/@gannonh/agent-pm)
AgentPM is a planning and orchestration system for AI-driven software development. Installed locally as an MCP server, it integrates with any IDE that supports Anthropic's Model Context Protocol specification, including Cursor, Augment, VS Code Copilot, Cline and Roo.
AgentPM serves the role of product manager, helping developers plan, prioritize, and execute complex projects:
- Develops comprehensive requirements
- Breaks down complex projects into actionable tasks with clear dependencies
- Orchestrates implementation with context-aware assistance
- Delivers relevant documentation and context when needed
- Guides technical decisions and systems design
- Promote software development best practices (TDD, vertical slicing)
Why AgentPM?
- Frictionless Setup: Get started simply by chatting with your coding agent - no CLIs or complex rules needed.
- Token/Context Optimization: Consolidates functionality around a core set of dynamic tools that are contextually economical and easy for coding agents to use and understand.
- Intelligent Context Management: Delivers the right information to your coding agent at the right time, optimizing token usage and eliminating the need for manual context handling or "memory banks".
- Structured Output: Automatically generates clear, human-readable markdown documents; no need to decipher JSON or plain text files.
- Integrated Documentation Retrieval: Automatically retrieve relevant documentation through Context7 integration.
- Comprehensive Task Management: Creates well-structured tasks with proper dependencies, priorities, implementation details, and status tracking. Complex work can be broken down into manageable subtasks with clear relationships.
- Flexible Requirements Process: Works with or w
342ef562575fOBSERVED · 2026-10-08Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add agent-pm --env ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} --env ANTHROPIC_MAX_TOKENS=${ANTHROPIC_MAX_TOKENS} --env MAX_TOKENS=${MAX_TOKENS} --env PERPLEXITY_API_KEY=${PERPLEXITY_API_KEY} -- npx -y @gannonh/[email protected]{
"mcpServers": {
"agent-pm": {
"command": "npx",
"args": [
"-y",
"@gannonh/[email protected]"
],
"env": {
"ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
"ANTHROPIC_MAX_TOKENS": "${ANTHROPIC_MAX_TOKENS}",
"MAX_TOKENS": "${MAX_TOKENS}",
"PERPLEXITY_API_KEY": "${PERPLEXITY_API_KEY}"
}
}
}
}Exposed tools (22)
16 read · 4 write · 2 destructive. Blast radius: 2 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
Authentication | read | User login and registration |
Context7 | read | Retrieves up-to-date documentation and code examples for any library. |
Features | read | Key features of the project |
Frontend | read | User interface |
Unknown | read | Could not extract components |
apm_complexity | write | Analyze task complexity, generate expansion recommendations, and create reports |
apm_create_session | write | Create a new session with optional initial context |
apm_dependencies | read | Manage task dependencies |
apm_destroy_session | destructive | Destroy a session by ID |
apm_get_session | read | Get a session by ID |
apm_project_brief_create | write | Create a project brief through an interactive interview process and generate tasks. IMPORTANT: Task generation is a long-running operation that may take several minutes to complete. Always inform users to be patient when this operation is in progress. |
apm_project_brief_result | read | Get the result of a completed project brief interview operation. Call this after apm_project_brief_status indicates the operation is complete. If the operation is still running, inform users to be patient as task generation can take several minutes. |
apm_project_brief_status | read | Get the status of a project brief interview operation. IMPORTANT: If the status is |
apm_task | read | Manage and query tasks in the project |
apm_task_generate | read | Generates individual task files in ${ARTIFACTS_DIR} directory based on ${ARTIFACTS_FILE} |
apm_task_modify | destructive | Create, update, and delete tasks and subtasks |
apm_update_session | write | Update a session |
context7_library_docs | read | Fetches up-to-date documentation for a library. You must call |
context7_library_id | read | Required first step: Resolves a general package name into a Context7-compatible library ID. Must be called before using |
get-library-docs | read | Fetches up-to-date documentation for a library. You must call |
resolve-library-id | read | Required first step: Resolves a general package name into a Context7-compatible library ID. Must be called before using |
test-tool | read | A test tool |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (5 observation(s))
- Network
- declared (2 observation(s))
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (8)
apm_destroy_session, apm_task_modify
.prettierignore
import { PROJECT_ROOT } from '../../config.js';const toolModule = (await import(`../../../src/mcp/tools/${toolName}.js`)) as ToolModule;import { ARTIFACTS_DIR, ARTIFACTS_FILE } from '../../config.js';import { registerTaskModifyTool } from '../../mcp/tools/task-modify/index.js';import { registerDependenciesTool } from '../../mcp/tools/dependencies/index.js';@anthropic-ai/sdk, @modelcontextprotocol/sdk, dotenv, lru-cache, openai, ora, uuid, zod
Gates applied: no_behavioural_pass.
342ef562575ffull audit observations/trust-audit/mcp-server/gannonh__agentpm.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-08 | 342ef562575f | SAFE | B | 89 | first audit |
Questions
What is the AgentPM MCP server?
MCP server for the planning and execution of AI-assisted development projects.
What tools does AgentPM expose?
22 in total: 16 read-only, 4 that write, and 2 that can delete or overwrite (apm_destroy_session, apm_task_modify). Every one is listed on this page with its risk.
Is AgentPM safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 2 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does AgentPM need?
It reads ANTHROPIC_API_KEY, ANTHROPIC_MAX_TOKENS, MAX_TOKENS, PERPLEXITY_API_KEY and PERPLEXITY_MAX_TOKENS from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does AgentPM run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as @gannonh/agent-pm at 0.1.4.
How current is this page?
The grade is for one exact copy of the source (342ef562575f), read on 2026-10-08. The repository is watched and re-audited when it changes.