Skill Claude CodeCodex
Meta — seal a finished engagement by moving its folder into context/archive/ and stamping engagement.md status: closed. Requires a delivered report/final-report.md. Use AFTER /pt-report. Trigger phrases: "archive engagement", "close the engagement", "seal it", "zarchiwizuj test", "zamknij engagement".
Skill Claude CodeCodex
Stage 3 (optional, white/grey-box) — enumerate the repositories the target is built from (the R-NNN registry) + authoritatively determine each repo's tech stack from the code, then build the security-focused code map (auth, crypto, input handling, sinks/sources, cross-repo trust boundaries, each code-routed…
Skill Claude CodeCodex
Stage 3 (optional) — plan the code analysis: bind each code-routed objective to the methodology right for the repo's class+stack and decompose it into concrete analysis passes (CA-NNN), producing code/code-plan.md per code-plan-schema.md. The input runbook the agentic scanner executes. Use AFTER /pt-code-map, BEFORE…
Skill Claude CodeCodex
Stage 3 (optional) — targeted secure code review: for each code-plan pass, trace data flows source→sink (incl. cross-repo), confirm or refute each code-routed objective, triage the agentic scan candidates, and file findings (tier Potential, one file each, repo-qualified evidence) per finding-schema.md while updating…
Skill Claude CodeCodex
Stage 3 (optional) — execute the code-plan's analysis passes as an AI agent (agent-first; optionally using available tools — SAST/SCA/secret — when they add value over pure reasoning), producing candidate findings under code/scan-results/ /. Use AFTER /pt-code-plan, BEFORE /pt-code-review. Trigger phrases: "scan the…
Skill Claude CodeCodex
Stage 4 — ingest the executor-provided evidence for each dynamic test case, assign a verdict, and map it to a finding action + ledger status per test-objectives-schema §5: file Verified / Potential-Unverifiable findings, promote code Potentials, or append follow-up cases (the re-plan loop). Use AFTER the cases are run…
Skill Claude CodeCodex
Stage 4 — build the dynamic test runbook the human executes: per in-scope objective (and code Potential to verify), concrete tool + exact command + safety class + evidence path, per dynamic-test-plan-schema.md. Tool selection by target class is where technology-independence is realised. --refresh re-surveys the…
Skill Claude CodeCodex
Stage 4 — the LOCAL EXECUTOR. Runs the /pt-dynamic-plan runbook from this machine (the alternative to the human running the tools), resolving each case's tool via environment/tool-registry.md to a local binary / Docker container / MCP call, enforcing scope + the auto-exec ceiling, and writing the identical…
Skill Claude CodeCodex
Initialize the context/ workspace for 10x-pentest — scaffold context/{engagements,archive,foundation,environment}/ + READMEs (+ .gitignore for the machine-scoped environment/) and verify the shipped foundation contracts. Use BEFORE /pt-new when setting up a fresh workspace. Trigger phrases: "init pentest workspace"…
Skill Claude CodeCodex
Open a new engagement folder under context/engagements/ / with an engagement.md identity file (per engagement-schema.md); --from opens a retest seeded from a prior engagement. Use AFTER /pt-init, BEFORE /pt-scope. Trigger phrases: "new engagement", "start a pentest", "open engagement", "retest", "nowy engagement"…
Skill Claude CodeCodex
Stage 5 — the hard quality gate. Audit every finding (evidence-grounded, tiers correct, CVSS consistent, traceability resolves) and the coverage ledger (all objectives terminal, constraints satisfied), then write review/findings-review.md with a machine-readable verdict (pass|fail|waived). A FAIL blocks /pt-report.…
Skill Claude CodeCodex
Independent quality review of Stage 1 — audits target-profile.md + attack-surface.md for completeness (independently hunting un-enumerated surface), classification correctness, stack fidelity, traceability, and confidence honesty; verdicts + findings + interactive triage. Advisory, not a hard gate. Use AFTER…
Skill Claude CodeCodex
Stage 1 — build the target profile and enumerate the attack surface, producing target-profile.md + attack-surface.md per target-profile-schema.md. Classifies the target class(es) (the technology-independence pivot) and originates the S-/TB- stable IDs. Use AFTER /pt-scope, BEFORE /pt-threat-model. Trigger phrases…
Skill Claude CodeCodex
Stage 5 — aggregate the engagement into a client-ready report (report/final-report.md) plus machine-readable exports (SARIF 2.1.0 + JSON + CSV) and a coverage matrix computed from the ledger, per report-template.md. Deterministic — a rendering of the on-disk artifacts, gated on a passed /pt-pentest-review. Use AFTER…
Skill Claude CodeCodex
Meta — stateful engagement conductor. Re-derives the current stage from which artifacts exist on disk and routes to the next skill; --status prints the stage table without doing work. The "where am I / what's next" helper for an engagement. Trigger phrases: "where am I", "what's next", "engagement status", "conduct…
Skill Claude CodeCodex
Independent quality review of Stage 0 — audits scope.md for authorization integrity, coherence, testability, and do-no-harm soundness, then verdicts (SOUND/REVISE/RETHINK) + findings + interactive triage. Advisory, not a hard gate; the producer's readiness gate stays the floor. Use AFTER /pt-scope, BEFORE /pt-recon.…
Skill Claude CodeCodex
Stage 0 — interactively define an engagement's scope, Rules of Engagement, and authorization, producing scope.md per scope-schema.md and computing its three-tier readiness gate. The blocking input to every later stage. Use AFTER /pt-new, BEFORE /pt-recon. Trigger phrases: "define scope", "set scope", "rules of…
Skill Claude CodeCodex
Independent quality review of Stage 2 — audits threat-model.md + test-objectives.md for coverage completeness (independently hunting missed attack paths), abuse-lens sufficiency, prioritization soundness, objective quality, and ledger integrity; verdicts + findings + interactive triage. Advisory, not a hard gate. Use…
Skill Claude CodeCodex
Stage 2 — decompose the target (STRIDE per trust boundary + attack trees + MITRE ATT&CK) into threat-model.md, and derive the prioritized objective register test-objectives.md (the living coverage ledger) per test-objectives-schema.md. Use AFTER /pt-recon, BEFORE /pt-code-map (or /pt-dynamic-plan). Trigger phrases…
Skill Claude CodeCodex
Meta — configure the machine's toolbench for local Stage-4 execution. Discovers/records which security tools are available here and HOW to invoke each (local binary / Docker container / MCP server) + a per-tool auto-exec/safety policy, into environment/tool-registry.md per tool-registry-schema.md. Machine-scoped (not…