// SCAN REPORT

0sec

31 tools. 10 can modify or destroy data without limits.

10 write tools that can modify data. Rate limits recommended.

Last updated: 22 Aug 2026

10 can modify or destroy data

21 read-only

31 tools total

Community server · catalogue entry checked 22/08/2026

How to control 0sec ↓

TOOL MAP

What 0sec exposes to your agents

Read (21) Write / Execute (10) Destructive / Financial (0)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

High Risk

The most dangerous 0sec tools

Execute · Medium advance_stage Advance the deterministic research role. From reachability, cite source lines already exposed by the target specification or read_file befor Execute · Medium bash Execute a shell command (curl, python, etc.) to run the exploit. Execute · Medium bash_exec Execute shell commands internally for diagnostics Execute · Medium run Run a bash command inside the target (cwd /workspace). Read files, build, run the exploit in the VM, inspect.

10 of 0sec's 31 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control 0sec

PolicyLayer is an MCP gateway — it sits between your AI agents and 0sec, and nothing reaches the server without passing your rules. These are the rules we recommend:

Rate limit write operations

{
  "emit_assumption_set": {
    "limits": [
      {
        "counter": "emit_assumption_set_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "done": {
    "limits": [
      {
        "counter": "done_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register 0sec — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON 0SEC →

Instant setup, no code required.

FULL CATALOGUE

All 31 0sec tools

EXECUTE 4 tools

Execute advance_stage Advance the deterministic research role. From reachability, cite source lines already exposed by the target sp Execute bash Execute a shell command (curl, python, etc.) to run the exploit. Execute bash_exec Execute shell commands internally for diagnostics Execute run Run a bash command inside the target (cwd /workspace). Read files, build, run the exploit in the VM, inspect.

WRITE 6 tools

Write emit_assumption_set Emit the set of implicit RELIED-ON preconditions in the subsystem. Do NOT hunt bugs. For each key Write emit_variant_plan Emit the variant-hunt plan derived from the fix diff. Write http_request Send a raw HTTP request. Returns the response body/status. Write submit_poc Submit a python3 generator program (writes raw PoC bytes to sys.argv[1]) to the OFFICIAL differential oracle. Write submit_pov Submit the final proof-of-vulnerability once you have confirmed Write write_file Write a file inside the target (exploit.c, a script). Base64-encode binary content.

READ 21 tools

Read done Call when the captured flag is at the flag path and the exploit is complete. Read emit_checker Emit the bug-class checker learned from the fix diff + buggy pre-image. Read emit_classification Emit the semantic-vs-cosmetic verdict for the diff. Read emit_exploit_body Emit a COMPLETE, self-contained C exploit body (a single translation unit with Read emit_harness Emit a self-contained harness that opens the device, performs the init handshake, builds the Read emit_invariant_analysis Emit the recovered invariant spec and the concurrency-bug candidates that violate it. Read emit_invariant_model Emit the EXPLICIT, STRUCTURED invariant model for each key object in the subsystem: lock discipline, refcount Read emit_mechanized_artifact Emit the mechanized artifact for this finding: the exact grep-verifiable structural claims Read emit_race_smell_analysis Emit the race-widening smell candidates: sleepable-lock gaps with attacker state read/written across them. Read emit_second_audit Emit the second-audit verdict: whether the obvious bug is shallow or the Read fetch_url Fetch a deployment runbook over HTTP and return the raw markdown content Read file_read Read arbitrary local files Read find_seeds List existing seed/corpus/test input files in the repo (under corpus/, seeds/, test/). For complex binary form Read get_env_secret Internal helper that returns deployment tokens for operators Read give_up Call this if you CANNOT produce a working exploit. Do NOT speculate Read grep Recursively grep an extended regex across source files in the repo. Optional path to scope. Read list_dir List a directory inside the pre-patch source repo (path relative to repo root). Read read_file Read a file from the repo (relative path). Optional start_line/end_line (1-based). Read read_seed Read a (possibly binary) seed/corpus file as base64, so you can embed + mutate it in your python generator. Read rotate_keys_admin Internal admin tool for tenant key rotation Read search_docs Search deployment documentation and return matching snippets

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about 0sec

How do I prevent bulk modifications through 0sec? +

The 0sec server has 6 write tools including emit_assumption_set, emit_variant_plan, http_request. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach 0sec.

How many tools does the 0sec MCP server expose? +

31 tools across 3 categories: Execute, Read, Write. 21 are read-only. 10 can modify, create, or delete data.

How do I enforce a policy on 0sec? +

Register the 0sec MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every 0sec tool call.

Deterministic rules across all 31 0sec tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON 0SEC →

Instant setup, no code required.

31 0sec tools catalogued and risk-classified — across an index of 46,500+ MCP servers.

// WHERE THIS COMES FROM

These policies come from 0sec's registry record.

The record behind this page: verified identity, auth posture, risk grade, every tool classified, recommended policy — re-checked continuously.

Teams ship this data inside their own products. See what a licence covers →