Intercept security for Claude Code — deep AI security scans that supplement your scanners and write findings back to the Intercept platform. OAuth 2.1, no API key.
Skill Claude CodeCodex
Pull a repo's Intercept AI Advisors and verify each recommendation against the actual local code and tool state, emitting a per-rec verdict (already-mitigated / valid / overstated-FP / partially-done) with concrete local evidence. This is the plugin's local-verification superpower against advisor noise (e.g. catching…
Skill Claude CodeCodex
Natural-language security hunt across the repo — decompose a user-described hunt (auth bypass, tenant isolation, SSRF, insecure deserialization, business-logic) into targeted code reading, then write findings back to Intercept. Use for /intercept:hunt or when the user asks to hunt for a specific class of…
Skill Claude CodeCodex
Fix a specific Intercept finding end-to-end — verify it's real, propose and apply the code fix, run available tests, mark it resolved, and optionally trigger a rescan. Use for /intercept:fix or when the user asks to fix or remediate a specific finding.
Skill Claude CodeCodex
Establish what's security-relevant about this repo before an Intercept scan or hunt — detect the stack from the code, ask the platform (getrepocontext) what it already knows, and confirm only the gaps it can't infer. Never writes anything into the repo. Use at the start of /intercept:scan and /intercept:hunt, or when…
Skill Claude CodeCodex
Perform a deep, systematic AI security review of the whole application — security architecture and threat model, authentication and authorization, injection and input handling, cryptographic weaknesses, container images, live cloud and infra, CI/CD and supply chain, and business logic — by actively investigating what…
Skill Claude CodeCodex
Verify open Intercept findings by reading the code and following the data/taint flow, classify each as true/false/unknown, and write the verdict back via updatefindingstatus. Works on both scanner and AI findings. Use for /intercept:triage or when the user asks to verify, triage, or clear false positives.
Skill Claude CodeCodex
Foundation for working with the Intercept MCP server — auth model, the tool catalog and when to use each, pagination, error handling, the resolutionid-vs-id rule, the category enum mapping, and the security guardrails. Load this whenever any other Intercept skill activates, or whenever the user asks to scan, triage…