| name | offensive-osint |
| description | Router for the Offensive OSINT arsenal. Dispatches to focused sub-skills by task type. Covers the full external red-team surface: asset discovery, web enumeration, identity/SSO, secrets/dorks, post-credential workflows, cloud/infra, people/breach intel, and analysis/reporting. Companion to osint-methodology. Use for any authorized external recon, bug bounty, or ASM engagement. |
| version | 2.1.1 |
| triggers | ["external recon","external red team","red team external","attack surface management","ASM","bug bounty recon","bug bounty","reconnaissance","footprinting","asset discovery","start recon","new target","pick up recon","continue recon","assess target"] |
Offensive OSINT — Arsenal Router
Companion: osint-methodology — pipeline stages, triage rules, severity rubric. Load it alongside this router at the start of every session.
Scope gate: Authorized targets only. When scope is unclear, ask once before proceeding.
BEHAVIORAL CONTRACT
When triggered: Any external recon, bug bounty, ASM engagement, or general "where do I start" offensive OSINT request.
Execute:
- Load
osint-methodology — identify the pipeline stage and scope.
- Match the current task to a sub-skill using the sub-skill map below.
- Load that sub-skill and begin execution immediately.
- Propose the next concrete action without waiting to be asked.
- When one sub-skill's work completes, chain to the next relevant sub-skill autonomously — follow the pipeline priority order from
osint-methodology §7.1.
Output: Delegation to the appropriate sub-skill(s). This router produces no findings itself.
Gating rules: Authorized targets only. When scope is unclear, ask once before proceeding. Hard rules (below) are always-on across all sub-skills.
Chain to: Autonomously chain through sub-skills following osint-methodology §7.1 priority order: breaches → GitHub recon → misconfig sweep → cloud buckets → ports → email OSINT → web tech → Wayback → DNS/email security → certs/TLS → ASN/reverse DNS → typosquats.
Sub-skill map
Load the sub-skill that matches the current task. Each is self-contained and under 500 lines.
| Task | Sub-skill to load |
|---|
| Subdomains, ASN/BGP, DNS, CT, WHOIS/RDAP, wordlists | recon-asset-discovery |
| Web surface: Swagger/GraphQL paths, curl probes, Wayback, Postman, endpoint scoring | web-surface |
| IdP fingerprinting, Entra/Okta/ADFS/SAML, M365 deep enum, LinkedIn employee enum | identity-fabric |
| Secret regexes, dork corpus, GitHub code-search dorks, read-only validators | secrets-and-dorks |
| Post-credential: JWT triage, AWS IAM enum, GitHub scope enum, Slack workspace enum | post-discovery |
| Cloud-native fingerprints, K8s/container, CI/CD exposure, infra OSINT | cloud-and-infra |
| Username/email/phone, breach data, HudsonRock, Slack/Discord/Telegram, package registries | people-breach-intel |
| Scoring rubrics, attack-path hints, severity matrix, AI-assisted OSINT, archiving | analysis-and-reporting |
| Report generation: bug-bounty submission, client deliverable, vulnerability report | report-template |
Session start checklist
- Load
osint-methodology — identify the pipeline stage and scope.
- Load the sub-skill matching the current task from the map above.
- Propose the next concrete action without waiting to be asked.
Hard rules (always-on, all sub-skills)
- Never report a bypass without demonstrated impact.
- Run read-only validators (
secrets-and-dorks) before escalating any credential finding.
post-discovery is gated — confirm read-only validation passes first.- No destructive probes. No active scanning outside explicit written scope.
Structured Outrider run contract
Follow the shared run-contract instructions in ../_shared/run-contract.md.
- Contract skill identifier:
offensive-osint.
- Consume
skill_request version 1 and produce skill_result version 1 when participating in an Outrider run.
- Use evidence IDs for all claims; do not cite unregistered local paths as claim evidence.
- Discoveries are observations and do not expand scope or approval.
- Do not claim final finding validation; use
finding_candidate only when a human-reviewed candidate should be handed off.
- Do not directly edit
manifest.json, scope.yaml, run.jsonl, evidence.jsonl, or approvals.jsonl.
- Use policy-gated MCP with the explicit
run_dir; the Python control layer and MCP boundary must reevaluate current controls.
- Router role: this skill may select another shipped skill while retaining the same
run_id and request context.
- Router role: do not invent unknown skill names, bypass contract validation, or aggregate unsupported narrative into a validated finding.
