google/mantis is a modular set of security-review skills that coding agents use to find, reproduce, and patch software vulnerabilities. It is for adapting automated security testing to application, infrastructure, machine-learning, hardware, and firmware projects, with expert review required for results. The catalogue skills and instruction provide the workflow for using Mantis with coding agents.
Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/google/mantisnpx agentmods add skills/google/mantis/mantis-reproduceWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/google/mantis/mantis-reproduce)<a href="https://agentmods.dev/skills/google/mantis/mantis-reproduce"><img src="https://agentmods.dev/badge/skills/google/mantis/mantis-reproduce.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00047 | $0.11129 |
| Opus 5 | $0.00023 | $0.05565 |
| Sonnet 5 | $0.00009 | $0.02226 |
| Haiku 4.5 | $0.00005 | $0.01113 |
Grade A, and why
mantis-reproduce scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 808 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Reproducer (/mantis-reproduce)
System Goal
Integration Test Engineer. Designs crash reproducers or inputs and executes them inside isolated sandbox environments to empirically verify bugs.
Command Definition
- Command:
/mantis-reproduce [--reattack] [--finding_id=<uuid>] [--force] [--target_root=<path>] [--state_root=<path>] [--snapshot_root=<path>] [--snapshot_id=<SNAPSHOT_ID>] [--snapshot_pinned=<true|false>] - Description: Generates and runs crash reproducers to verify security flaws.
- Parameters:
--reattack: When executing as part of patch verification to isolate re-attack outcomes.--finding_id: The specific finding UUID to reproduce. Must be provided and is required when--reattackis specified.--force: Override/bypass eligibility checks for targeted normal runs.--target_root: Path to the root of the target codebase under test (defaults to.). AUTHORITATIVE when supplied — overrides--snapshot_root(Block A step 1a); the sentinel check is skipped for this tree (e.g. a patched shadow during re-attack verification).--state_root: Path to the root of the Mantis state directory containingworkspace/(defaults to.).--snapshot_root: Root of the pinned immutable code snapshot for this pass. Consumed by Block A (Step 0) when--target_rootis not supplied.--snapshot_id: The SNAPSHOT_ID string of the pinned snapshot, consumed by Block A (sentinel) and Block B (snapshot match check) in Step 0.--snapshot_pinned: Whenfalse(set bymantis-patchduring re-attack on a patched shadow), the reproduce sub-agent MUST skip the snapshot sentinel/match check for this invocation — the--target_roottree is authoritative and sentinel-exempt (Block A step 1a).
Input/Output Contract
- Reads:
state_root/workspace/findings/(viable/conditional findings).target_root/(Repository source files to analyze trigger paths).state_root/workspace/archive/.repro_attempts.json.state_root/workspace/.mantis_state.json(to track current loop pass).
- Writes:
- PoC reproduction files (e.g.
poc_[uuid].pyorcrash_[uuid].payloadinsidestate_root/workspace/reproducers/). - If run normally: updates findings in-place under
state_root/workspace/findings/(sets"repro_status","repro_file_path","run_command","repro_output", and appends history). Updates status to"VALID"if provisionally valid. - If run with
--reattack: updates findings in-place understate_root/workspace/findings/(sets"reattack_status","reattack_file_path","reattack_run_command","reattack_output","reattack_variants", and appends history with stage"reattack"). Does not modify"repro_*"fields or"status". Exception: may atomically downgradepatch_statusper INV-1 in Step 6 (never persistVERIFIED_SECUREalongside a non-failed_to_bypassreattack_status). - Updates
state_root/workspace/archive/.repro_attempts.jsonatomically. - Stamps
"repro_snapshot_id"/"reattack_snapshot_id"on updated findings and stores.repro_attempts.jsonvalues as{count,last_snapshot}objects (bare integers still read correctly).
- PoC reproduction files (e.g.
- Preconditions:
- Findings must exist in
state_root/workspace/findings/. - Sandbox/container runtime environment must be available.
- Findings must exist in
- Idempotency Guarantee:
- Updates findings in place. Uses
state_root/workspace/archive/.repro_attempts.lockfile locking and atomic temporary file swaps (os.replaceonstate_root/workspace/archive/.repro_attempts.json.tmp) to guarantee concurrency safety and retry stability. - Snapshot-aware: regenerates the PoC when the finding's snapshot no longer
matches; refuses to emit a negative verdict without reached-sink evidence.
Re-attack verdicts on a snapshot mismatch are governed by the C5
unpatched-baseline re-run (Step 6), which supersedes the legacy blanket
refusal — a
failed_to_bypassverdict is only written after C5 has confirmed the unpatched baseline still triggers on the current snapshot.
- Updates findings in place. Uses
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 808 lines · 47 tokens per session scan A b5b704b9cbc8
mantis-reproduce is a skill published in the GitHub repository google/mantis (943 stars, last pushed 3d ago), licensed Apache-2.0. It adds 47 tokens to every session and 11,129 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
cc-python-dev
Python 开发规范,适用于类型注解、验证模型、pytest 和 uv 工具链;不负责 review、debug 或运维流程。.
run-tests
Run tests matching a pattern. Use when user says "test", "run tests", or asks to verify changes.
eval-harness
Professional Eval Harness Expert skill. Integrate LLM API workflows, safe system prompt guidelines, and agentic workflows.
systematic-debugging
4-phase root cause debugging: understand bugs before fixing.
react-patterns
React 18/19 patterns including hooks discipline, server/client component boundaries, Suspense + error boundaries, form actions, data fetching, state management decision trees, and accessibility-first composition. Use when writing or reviewing React components.
golang-testing
Go testing best practices including table-driven tests, test helpers, benchmarking, race detection, coverage analysis, and integration testing patterns. Use when writing or improving Go tests.