Shadow-Core Sentinel
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Shadow-Core SentinelDid my edit to main.py land?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Shadow-Core Sentinel — MCP Filesystem Telemetry
Sentinel records every file change under a watched directory, with a SHA-256 of each file, so a change can be confirmed against what actually happened on disk rather than assumed.
It records what changed, never whether the change is correct. It is a filesystem oracle, not a semantic one — it will not catch a deleted function, a wrong value, or a broken test. Linters and tests remain the tools for that.
Key features
Non-blocking — hashing is offloaded to a thread pool so OS events are not dropped while a large file is read.
Cryptographically verified — every event carries a SHA-256, so "did this file revert to its original state?" is answerable.
Context-optimised —
recent_changesanswers "did my edit land?" in tens of rows.query_eventstakes a whole date, which on a busy project is tens of thousands (measured: 19,936 in one day).Multi-project — several projects watched at once, each with its own database and audit directory. Adding a watch never removes another, so two sessions cannot silently stop each other's monitoring.
Idle suspension with gap recovery — an inactive project suspends rather than being watched forever. Suspension is not removal: history stays intact and the next prompt resumes it. Changes made while suspended are reconstructed from a SHA-256 comparison and written into the trail marked as detected-on-resume, so the unwatched period is visible rather than missing.
Atomic-write aware — editors that write via a temp file and rename record as a single
MODIFIEDof the real path, with no phantomDELETEof the file they just replaced.Noise guard — high-churn directories (
node_modules,.git,venv, build output) are ignored by default.
Related MCP server: trueline-mcp
Requirements
Python 3.11. The pinned versions in requirements.txt are verified against
it.
Install
pip install -r requirements.txtFor development — this is what you need to run the test suite, which
requirements.txt alone does not provide:
pip install -e ".[dev]"Run
python main.pySentinel boots watching nothing and stays idle until a session calls
watch_project. That is deliberate: it starts with the machine, and coming up
recording a directory nobody asked about means CPU spent hashing and an audit
trail nobody reads.
To start it automatically at logon, see INSTALL.md.
Flag | Default | Meaning |
|
| MCP SSE endpoint |
|
| Bind address — loopback is deliberate |
|
| HTML dashboard |
| off | Run without the dashboard |
| none | Watch a directory at startup |
|
|
|
--port is a deprecated alias for --dashboard-port.
Environment variables
Variable | Default | Meaning |
|
| MCP SSE endpoint |
|
| Dashboard port |
|
| Set |
|
| Where per-project audit data is written |
|
| Startup watch directory |
|
| Deletes all recorded audit data at startup. See below |
|
| Idle time before a watch suspends; |
|
| How often idle watches are checked |
|
| Logging level |
Flush on start
SENTINEL_FLUSH_ON_START defaults to true: every start deletes all
recorded audit data, so a run begins with no history. This completes what
Sentinel already did — it boots watching nothing, and its in-RAM ring starts
empty — and it bounds disk, which nothing else did. Before this the trail had
reached 252 MB with no retention policy at all, 38% of it a dead project
created by a typo in a watched path.
What you lose, stated plainly: cross-restart forensics. list_audit_dates,
get_daily_report and query_events can only answer about the current run,
and "what changed while I wasn't looking" — the one question git cannot answer
— is unanswerable across a restart, because the evidence is deleted first. Gap
reconstruction still works, but only across a suspend/resume inside one run.
Set SENTINEL_FLUSH_ON_START=false to keep history.
The flush only removes things it recognises as its own: a directory holding a
sentinel.db or Sentinel's markdown artifacts, an empty project directory, or
a loose sentinel.*/audit-*.md/snapshot-*.md/gap-*.md at the audit root.
Anything else is left alone and logged, and an AUDIT_DIR closer than three
path components to a filesystem root is refused outright — a mis-set
AUDIT_DIR must not be able to delete source.
MCP client configuration
Sentinel speaks MCP over SSE, not stdio. It is not spawned by the client — it must already be running, and the client connects to it:
{
"mcpServers": {
"shadow-core-sentinel": {
"type": "sse",
"url": "http://127.0.0.1:7702/sse"
}
}
}A "command"/"args" entry — the stdio spawn form — does not work here.
The client launches the process, waits for stdio that never comes, and hangs,
because main.py runs mcp.run(transport="sse", ...) and serves HTTP instead.
Verify it is up:
curl http://127.0.0.1:7702/healthUsing it
At the start of a session:
watch_project(path="<absolute path of the working directory>")Additive and idempotent — it never stops another session's watch, and re-calling it for an already-watched directory only renews its lease.
Before reporting that a change is complete:
recent_changes(minutes=15)Compare files you intended to change against what the filesystem recorded. This catches an edit that silently did not land, and files changed that were not meant to be touched. It is worth most after a build, install, or generated-file step, where an exit code of 0 is not evidence that a file was written.
Endpoints
Endpoint | Purpose |
| Liveness, watch list, and |
| Keepalive; renews and resumes the watch for a path (localhost only) |
| Graceful stop without elevated |
| Dashboard, one tab per watched project |
Selecting a dashboard tab is a client-side view change: it does not move the server's default project or affect another session.
Tests
python -m pytest -m "not slow"Layout
File | Responsibility |
| Startup: build state, wire components, run |
| The nine MCP tools and two resources |
| Which project a dashboard request is answered from |
| Dashboard HTTP layer and template |
|
|
| watchdog handler: ignore, debounce, atomic-write handling |
| The watched projects and longest-prefix event routing |
| Idle suspension, resume, and gap reconstruction |
| SQLite event store, one per project |
| Markdown audit logs and snapshots |
| Settings and the ignore filter |
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- Alicense-qualityCmaintenanceA production-oriented MCP server for coding agents that enables multi-project management through secure file operations, Git integration, and safe command execution. It supports project discovery across multiple root directories and provides robust audit logging with both STDIO and HTTP transport options.142MIT
- Alicense-qualityAmaintenanceMCP server providing hash-verified file editing and targeted reads, reducing context consumption and preventing silent corruption by requiring content hashes for edits.1134Apache 2.0
- AlicenseAqualityCmaintenanceMCP server for on-disk patching of binary artifacts with SHA-256 manifest, byte-level patch application, and manifest-driven restore, ensuring auditability and reversibility.4MIT
- Alicense-qualityCmaintenanceA tuned filesystem MCP server for Codex-style development, offering fast file operations, bounded output, and safe edits.MIT
Related MCP Connectors
A MCP server built for developers enabling Git based project management with project and personal…
Zero-install remote MCP server for proof-of-existence file attestation.
Personal MCP server for humans who create. Proof of authorship, license control.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/Shadow-Core-Omnipotence/Shadow-Core-Sentinal'
If you have feedback or need assistance with the MCP directory API, please join our Discord server