vestige
Think before doing, prove before saving. Vestige is a local, deterministic safety kernel for AI agents. Every memory write and agent action is recorded in Strata, an append-only log, so you can see exactly what your agent did and why.
No outside reports on file
Vestige helps you see what happened after it happens, but logging alone doesn't prevent an agent from making mistakes or being tricked—you still need to restrict what the agent can access and do.
Based on public adoption and transparency signals as of Oct 27, 2026. Our opinion, not a code audit or security guarantee.
What it can access
- Local filesNo local folder access notedLOW
- Command executionNo shell command runner notedLOW
- Network reachConnects to outside service APINOTED
- CredentialsNo API token requiredLOW
- Package sourcenpm (vestige-mcp-server)PUBLIC
Inferred from this listing’s public package details and description. Check the project’s source before installing.
Install
npx -y vestige-mcp-serverEverything known
We found nothing about this project in the sources we read (last checked 2026-10-30). That does not mean it has no problems.
0 incidents · 0 advisories · 0 news articles · 0 community threads · checked 2026-10-30
Popularity grade: C (from public signals such as stars and downloads, not a code review). Graded 2026-10-27.
Sources we read: OSV, GitHub Advisory Database, CISA Known Exploited Vulnerabilities, 44 security news feeds and our incident library. Last checked 2026-10-30.
About Vestige
Vestige is a safety tool that records everything an AI agent does and remembers in an append-only log called Strata. It lets you review exactly what actions the agent took and why, giving you a complete audit trail of the agent's behavior and decisions.
Who it’s for
Developers building AI agents who want to track agent actions for safety, debugging, and accountability. Teams that need to audit or review what their agents did before trusting them with important tasks.
Alternatives
- OpenAI Fine-tuning + Auditing: Use model fine-tuning with manual audit trails
- LangChain Callbacks: Track agent steps using LangChain's built-in callback system
- Anthropic Logs + Claude: Use Claude API logging to record requests and responses
- Custom audit middleware: Build your own logging layer around agent tool calls
Setup
Check these steps against the project’s own instructions before running anything.
Follow the project's README at https://github.com/samvallad33/vestige (opens external site) for installation and configuration instructions specific to your setup.
Why this grade
Signal-based grade (3/7 points): well starred on GitHub, public source code. This reflects adoption and transparency, not a code review.
Popularity and upkeep
- GitHub stars
- 658
- Weekly npm downloads
- 355
- Listing last refreshed
- 2026-10-29
These are public popularity signals, not a maintenance check. Look at the repository’s commit history and open issues.
Risks to watch for
- Append-only logs can grow large: Storage space fills up over time if logs aren't managed, potentially causing performance issues
- Log contents may expose sensitive data: Recorded agent actions and memories could contain API keys, personal information, or business secrets if the agent handles them
- False confidence in safety: Reviewing logs after problems occur doesn't prevent damage—auditing is useful but not a substitute for restricting what agents can do
- Local access is still access: Anyone with file system access to the log can read the agent's history and any sensitive information stored there
- Determinism relies on setup: The tool only records actions; if the agent is configured to connect to harmful services, logging won't stop it
How to lower the risk
Least-privilege permissions
- Give the agent only the permissions it absolutely needs for its job
- Restrict which files, APIs, and services the agent can reach
Network isolation
- Run the agent in a sandboxed environment with limited network access
- Don't let it make outbound requests unless required
Monitoring
- Regularly review the append-only logs for unexpected behavior
- Set alerts if the agent tries to access resources outside its normal patterns
Red flags to watch for
- Agent making requests to unknown services or URLs
- Sudden increases in data access or file operations
- Attempts to modify or delete earlier log entries
Questions
How do I install Vestige?
Run npx -y vestige-mcp-server in your terminal, or add the server entry to your claude_desktop_config.json or .cursor/mcp.json as shown in the Setup section on this page. Full setup docs and environment variables are in github.com/samvallad33/vestige.
Are there security reports about Vestige?
No reports found for this package in our research (last checked 2026-10-30). Vestige currently holds Grade C (Moderate risk signals) based on public adoption and transparency signals. Key risk areas to watch include append-only logs can grow large, log contents may expose sensitive data, false confidence in review.
Is Vestige actively maintained?
Vestige has 658 GitHub stars and 355 weekly npm downloads. Our directory data was last refreshed on 2026-10-29. Check recent commit activity, release tags, and open issues at github.com/samvallad33/vestige before pinning it in production.
What are alternatives to Vestige?
Notable alternatives to Vestige include OpenAI Fine-tuning + review (Use model fine-tuning with manual review trails), LangChain Callbacks (Track agent steps using LangChain's built-in callback system), Anthropic Logs + Claude (Use Claude API logging to record requests and responses), Custom review middleware (Build your own logging layer around agent tool calls). You can also compare Vestige side by side with other MCP servers in the Related tools section on this page.
README badge and team notes
Maintain or use Vestige? Show its live grade in your GitHub README, docs site, or run a check from the terminal before you install.
GitHub README Markdown:
[](https://agentarmorhq.com/skills/samvallad33-vestige)HTML embed:
<a href="https://agentarmorhq.com/skills/samvallad33-vestige"><img src="https://agentarmorhq.com/api/badge/samvallad33-vestige" alt="AgentArmorHQ Risk Grade for Vestige" height="20" /></a>Terminal check:
curl -fsSL https://agentarmorhq.com/api/cli | node --input-type=module - check "samvallad33-vestige"Share with your security lead
Need sign-off to use vestige? Copy this note for Slack, a ticket or a pull request.
### AgentArmorHQ Security Review Memo: vestige
- **Listing URL:** https://agentarmorhq.com/skills/samvallad33-vestige
- **Type:** App connector
- **Risk Grade:** Grade C (Moderate risk signals)
- **Outside Security Advisories on File:** 0
- **Public Adoption:** 658 GitHub stars, 355 weekly npm downloads
#### Permission & Blast-Radius Signals
- **Local files:** No local folder access noted
- **Command execution:** No shell command runner noted
- **Network reach:** Connects to outside service API
- **Credentials:** No API token required
- **Package source:** npm (vestige-mcp-server)
#### Recommended Hardened Setup
- **Command:** `npx -y --ignore-scripts vestige-mcp-server`
- **Controls:** Pin an exact version in `.mcp.json`, scope access to a single workspace folder, and use a least-privilege read-only token where supported.
_Based on public adoption and transparency signals as of Oct 27, 2026. Our opinion, not a code audit or security guarantee._Key facts
- Type
- App connector
- Publisher
- samvallad33
- Source code
- github.com/samvallad33/vestige
- Install command
- npm install vestige-mcp-server
- Est. context load
- ~750 tokens / turn (~4 tool schemas, 0.38% of 200k context)
- Popularity
- 658 GitHub stars, 355 weekly npm downloads
- Last updated
- 2026-10-29
- Last graded
- 2026-10-27
Community
Discussions for Vestige
Ask questions or share configuration tips. Governed by our Community Rules & Terms.
No posts match this filter yet.
Be the first to ask a question or share a working tip.
Check before you install, in Claude
Your assistant answers Block, Review or No known issues for any package before it installs it.
Get this check inside Claude