creativityResmi
Resmi Sağlayıcı Skill'iView repo
Emit AIDLC inception artifacts as GitHub issues: one EPIC per intent, one child per AIDLC unit, natively sub-issue-linked, each in the repo's mandatory plain-terms opening and five technical sections with deterministic acceptance. Delivery loop generation is a two-stage protocol: Step 7 composes loop drafts as branch artifacts (invoked in Run A after delivery-planning approval); Step 8 materializes the issues from approved drafts (invoked in Run B after loop-proposal approval). Orchestrators, ev
Dosyalar1 dosya
SKILL.md979 satır
Loading editor…
Kurulum
ÖnerilenTek komut — ajanınız otomatik olarak devreye alır.
Kurulum komutunu görmek için yukarıdan bir AI aracı seçin.
veya
Manuel Kurulum
Daha fazla adımDosyayı indirin ve ajanınızın sistem istemine yapıştırın.
Skill detayları
Versiyonv1.0.0
YazarAWS Labs
Kategoricreativity
Skill IDawslabs/agentic-developer-platform/modules/agent-factory/skills/aidlc-emit-issues
İlgili skill'ler
Codex BridgeDelegate a single bounded coding or code-review task to the OpenAI Codex CLI (runs against Bedrock), then review Codex's output yourself before committing. ONLY use this skill when the triggering issue or comment EXPLICITLY names Codex — e.g. "use Codex", "ask Codex to review this", "pass this to Codex", "have Codex write X". Never invoke it autonomously, speculatively, or because a task merely looks like something Codex could do. If nobody asked for Codex, do the work yourself and do not run thSkypilotUse for issue-driven GPU capacity requests: express workload requirements, let SkyPilot select and provision machines, and prepare those machines for Superplane EKS workloads through the hybrid networking and node-join path.Stage 1 TriageLightweight fingerprinting of a malware sample — hashes, file type, entropy, strings, candidate IOCs. Use this skill whenever you start a 7-stage malware analysis pipeline, need a quick file identification before deeper analysis, or want to check a sample against known-hash databases before committing compute. This is always the first stage and must run before Stage 2 (OSINT), Stage 3 (static), or Stage 4 (dynamic).Stage 2 OsintConsult OSINT sources and synthesize a prior hypothesis that steers downstream analysis. M1 scope is MalwareBazaar (no auth, no vault dependency) + web search + MITRE ATT&CK; VirusTotal + Shodan + urlscan are commented out in the skill until vault credentials are provisioned. Use this skill whenever you have a sample's hashes + candidate IOCs (from Stage 1 triage) and want to check reputation, find prior reporting, or decide whether deeper analysis is worth the compute. Always run after Stage 1 Stage 5 CorrelationCross-reference dynamic findings against threat intel (MISP, VT, web reporting) + MITRE ATT&CK to derive family + TTPs + campaign linkage. Use this skill whenever you have Stage 4 dynamic observations (C2 IPs, mutexes, injection targets) and want to link them to known campaigns, confirm Stage 2's hypothesis, or produce MITRE TTP IDs with specific per-stage evidence. Runs after Stage 4, or after Stage 3 if Stage 4 was unavailable — can work with partial data. Skip on Stage 2 short-circuit.Stage 6 VerdictSynthesize severity (malicious/suspicious/benign), confidence (high/medium/low), STIX 2.1 IOC bundle, CSV feed, and concrete recommended actions (host isolation, token revocation, perimeter blocks, fleet-wide hunt queries) from all prior stage envelopes. Use this skill whenever earlier stages have produced findings (or on a Stage 2 short-circuit to benign) and you need a machine-readable verdict that IR teams can ingest into Splunk/Elastic/Sentinel or block at the perimeter. Always runs — even s