Open-source, self-hosted AI penetration testing framework: maps your attack surface into a graph, autonomously exploits it from a Kali sandbox with human approval gates, and opens PRs that fix what it finds. MCP both ways: plug in any MCP server as a tool, or drive RedAmon from Claude Code or your own agent.
RedAmon is an AI framework that connects reconnaissance, exploitation, and post-exploitation activities into an automated security-testing pipeline, while organizing findings and proposing code fixes. It is for authorized penetration testing, security research, and education. The catalogue entries provide skills and instructions for working with the framework.
This repository also configures its own agents. See what redamon tells them →
Skill Claude CodeCodex
Adding a Community Agent Skill: a Markdown attack-workflow file that users import from the catalog, which then competes in the Intent Router and is injected into the agent's system prompt. No Python, no rebuild - but it must name only real tools and stay classifiable. Trigger: adding or editing a .md workflow under…
Skill Claude CodeCodex
Adding partial-recon support for a tool: running a single pipeline phase on demand from the workflow graph, reading its inputs from the existing Neo4j graph and merging results back. The input-node map decides the modal UI, and some input types must never be manually entered. Trigger: adding a tool to partial recon; a…
Skill Claude CodeCodex needs its repo
Wiring a new tool the AI agent can call (not the recon pipeline): the tool registry, the phase map, the hardcoded dispatch chokepoint, and the duplicated execution paths that make a tool work in single mode but silently break in parallel plans. Trigger: adding or editing a tool the agent invokes; editing…
Skill Claude CodeCodex
Adding a built-in Agent Skill (an attack technique like ssrf, xxe, rce) that ships hardcoded in RedAmon: classified by the Intent Router, injected into the agent prompt, toggled per project, badged in the chat drawer. Ten layers across agentic and webapp, two of which fail with no error. Trigger: adding or editing a…
Skill Claude CodeCodex
Writing to the Neo4j attack-surface graph in RedAmon: the tenant-isolation MERGE key every entity node must carry, where graph methods live (mixins, not the client), and the schema places that must be updated together. A MERGE missing the tenant key silently merges one project's data into another's. Trigger: editing…
Skill Claude CodeCodex
Adding an LLM provider to RedAmon: the credential boundary (keys must never reach scan containers), prefix-routed model ids, and the provider registry. A misrouted model id or a leaked key are the two failures this guards. Trigger: adding or editing an LLM provider; editing parsemodelprovider in…
Skill Claude CodeCodex
Spawning and hardening scan containers from the recon orchestrator: the security flags that look correct and break the container, and the sibling bind-mount path handling. capdrop and no-new-privileges were each reverted after breaking real scans. Trigger: editing reconorchestrator/containermanager.py; changing how a…
Skill Claude CodeCodex needs its repo
Changing or adding a project setting / default value in RedAmon. A single setting is duplicated across Prisma, two separate Python settings modules, the orchestrator defaults endpoint, and the frontend fallback; miss a layer and the UI shows one value while the backend uses another, and existing projects keep the old…
Skill Claude CodeCodex
Wiring an LLM into a recon tool's decisions ("let AI pick {feature} for {tool}"): the never-raise contract, the per-target cache, the full+partial coverage, and the two UI toggles bound to one field. A raising AI helper or an empty fallback silently breaks or disables a live scan. Trigger: adding or editing a recon AI…
Skill Claude CodeCodex
Adding a new tool to the recon pipeline: the enrichment-module contract and its isolated wrapper (the actual fan-out and test call path), graph completeness, and the preset catalog that silently strips unknown settings. Miss the isolated wrapper and the tool never runs in parallel; miss the catalog and AI presets drop…
Skill Claude CodeCodex needs its repo
How RedAmon tests actually run and how to author them: the per-file Docker gate, the unit/integration/live tiers, and the failure modes that make a green run a lie. Trigger: editing any test.py, .test.ts(x) or tests/.sh; a test that is red, skipped or xfailed; a request to "run the tests", "make it green" or check…
Skill Claude CodeCodex
Working on RedAmon's supply-chain scanner (offline OSV + GuardDog + retire + trufflehog): the offline OSV database that the scan path does not bootstrap, the world-readable requirement for the hardened scanner, and the soft-error markers that record what was never analysed. Trigger: editing…
Skill Claude CodeCodex
Working on the HTTP capture proxy and its replay/fuzz path: the egress guard that stops the proxy becoming an SSRF pivot, why it checks the resolved IP, and keeping capture off the scan's critical path. Trigger: editing scanners/captureproxy/ (captureaddon.py, egress.py, ingestworker.py); adding a forward/replay/fuzz…
Skill Claude CodeCodex needs its repo
Adding, removing or changing a tool on RedAmon's INBOUND MCP server, where external agents connect in with a personal access token. The wiki API reference is generated from the server's live tools/list and goes stale silently, and each tool's annotations and declared scopes are published to connected clients. Trigger…
Skill Claude CodeCodex needs its repo
The Priority Board's three-layer score model (rules BASE, REVIEW by the built-in AI or an MCP agent, a person's DECISION) and who may write which layer. A writer that sets a final score itself, or touches another layer, silently undoes a review or a person's verdict at the next run. Trigger: editing…
Skill Claude CodeCodex needs its repo
The plugin platform, always on: third-party plugin code runs isolated in the plugin runner, reaches RedAmon only through the Gateway, and changes agent behaviour only through text a person approved in core UI. The base repository holds no plugin logic, and a host that cannot run the platform is refused, never…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: