warden-aiBLOCK
The structurally-verified context layer for AI coding agents. Cut 50-90% of your agent's token spend.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
Verified context for AI coding agents.
Your AI agent burns tokens on noise. Warden stops that.
Less context. More signal. Verified.
Every token your agent spends on noise is a token it didn't spend on your actual problem. Warden fixes this.
Warden is a local context layer for AI coding agents. It drops into Claude Code, Cursor, Codex, Windsurf, Cline, Gemini, and 30+ other agents. Install once. Warden removes unnecessary tool-output noise, preserves important information, indexes your codebase for efficient retrieval, maintains useful memory across sessions — then verifies that compression did not lose critical information.
50-90% fewer tokens. Zero config. Zero cloud. Zero lock-in.
Benchmarked. 30-task suite. 72.3% overall reduction. 100% guard pass rate. 8ms avg overhead. [See the numbers →](
661c5fc483e0OBSERVED · 2026-10-12Connect
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.
claude mcp add warden-ai -- npx -y [email protected]
Exposed tools (40)
36 read · 4 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
code_review | read | Please review the following code carefully and provide detailed feedback. |
custom | read | Short desc. |
no_schema | read | No schema. |
ok | read | Short. |
ping | read | Returns pong. |
read_file | read | This tool reads a file from the filesystem. You can use it to read the contents of any file. Please note that the file path must be absolute. |
test | read | This tool reads a file from the filesystem. You can use it to read the contents of any file that you have permission to access. Please note that the file path must be absolute. |
warden_architecture | read | |
warden_call_graph | read | |
warden_ccr_status | read | |
warden_compress | read | |
warden_context_select | read | |
warden_dead_code | read | |
warden_file_read | read | |
warden_git_context | read | |
warden_grep | read | |
warden_handoff | read | |
warden_impact | read | |
warden_index | read | |
warden_memory_archive | read | |
warden_memory_failed_approaches | read | |
warden_memory_forget | read | |
warden_memory_list | read | |
warden_memory_mark_contested | read | |
warden_memory_reaffirm | read | |
warden_memory_recall | read | |
warden_memory_reject | read | |
warden_memory_save | write | |
warden_outcome_stats | read | |
warden_prune | read | |
warden_record_outcome | read | |
warden_report | read | |
warden_retrieve | read | |
warden_run_command | write | |
warden_run_tests | write | |
warden_search_symbols | read | |
warden_status | read | |
warden_sufficient_context | read | |
warden_task_report | read | |
write_file | write | Writes content to a file on the filesystem. You should provide the file path and the content to write. |
Trust audit
BLOCKgrade D · trust 69/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | FAIL |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (7 observation(s))
- Network
- declared (1 observation(s))
- Shell
- declared (4 observation(s))
- Dependencies
- not all pinned
- Secrets in source
- found
Findings (21)
if (lower === "id_rsa" || lower === "id_ed25519" || lower === ".npmrc")
if (lower === "credentials" || lower === ".netrc" || lower === ".pgpass")
writeFileSync(join(dir, "server.pem"), "-----BEGIN PRIVATE KEY-----\n");
writeFileSync(join(dir, "id_rsa"), "-----BEGIN OPENSSH PRIVATE KEY-----\n");
.clinerules
.continuerules
.mcprules
.prettierignore
.prettierrc.json
writeFileSync(join(dir, "id_rsa"), "-----BEGIN OPENSSH PRIVATE KEY-----\n");
expect(p).not.toMatch(/(^|\/)id_rsa$/);
import type { TaskContext } from "../../classifier/types.js";import type { TaskContext } from "../../classifier/types.js";import type { TaskContext } from "../../classifier/types.js";import type { TaskContext } from "../../classifier/types.js";import type { TaskContext } from "../../classifier/types.js";tree-sitter-wasms, web-tree-sitter, puppeteer, vitest
src/auth/middleware.ts:152: res.setHeader('Access-Control-Allow-Origin', process.env.CORS_ORIGIN || '*');src/auth/middleware.ts:155: res.setHeader('Access-Control-Allow-Credentials', 'true');Access-Control-Allow-Credentials was not being set when
curl -fsSL https://raw.githubusercontent.com/rynald0cst0ltziam/Warden-AI/main/install.sh | bash
Gates applied: no_behavioural_pass.
661c5fc483e0full audit observations/trust-audit/mcp-server/rynald0cst0ltziam__warden-ai.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 |
|---|---|---|---|---|---|
| 2026-10-12 | 661c5fc483e0 | BLOCK | D | 69 | first audit |
Questions
What is the warden-ai MCP server?
The structurally-verified context layer for AI coding agents. Cut 50-90% of your agent's token spend.
What tools does warden-ai expose?
40 in total: 36 read-only, 4 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is warden-ai safe to connect to an agent?
No — not without reading the findings first. The audit graded it D (69/100) and found 2 critical or high issues in the source. Each one is listed on this page with the file and line it is on.
What credentials does warden-ai need?
No credential environment variables were found in its source, so it appears to need none.
How does warden-ai run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as warden-ai at 0.7.1.
How current is this page?
The grade is for one exact copy of the source (661c5fc483e0), read on 2026-10-12. The repository is watched and re-audited when it changes.