01 Skill usage
QUICKSTART
from overmind.skills import AudioSkill
skill = AudioSkill.from_config()
result = skill.run({"bpm": 128.0, "energy": 0.85})
print(result["final_output"])
Pipeline composition
from overmind.compose import SkillPipeline
from overmind.skills import ResearchSkill, SecuritySkill
pipe = (SkillPipeline()
.pipe(ResearchSkill.from_config())
.gate(Confidence(min=0.7))
.pipe(SecuritySkill.from_config())
)
result = pipe.run({"topic": "Log4Shell risks"})
Memory-aware run
# Same subject re-uses past context automatically
result = skill.run({"query": "..."}, subject="project_x")
02 All skills
AudioSkill
Audio · Visualiser
Beat detection, mood classification, and preset selection for music visualisers. Takes audio feature vectors and returns visual direction parameters. Drives Spectraforge preset selection.
BPM trackingMood classificationPreset selectionSpectraforge integration
NetworkSkill
Security · Network
Network reconnaissance, vulnerability assessment, and risk synthesis. Processes scan outputs and structures findings into prioritised risk reports.
Recon analysisRisk scoringNmap / Shodan input
SecuritySkill
Security · Analysis
CVE triage, STRIDE threat modelling, and executive-level security reporting. Handles the analytical layer of security assessments.
CVE triageSTRIDEExecutive reports
HardwareSkill
Embedded · Firmware
Device analysis, firmware scaffolding, and embedded system debugging guidance. ESP32, Pico, and general microcontroller support.
ESP32 / PicoFirmware scaffoldingDebug analysis
CodeSkill
Code · Review
Code review, bug detection, test generation, and refactoring guidance. Language-agnostic — Python, C, C++, JavaScript, Zig.
ReviewDebugTest generationRefactor
ResearchSkill
Research · Synthesis
Source evaluation, structured synthesis, and research summaries. Handles multi-source ingestion and confidence-weighted output.
UniversalSkill
Pipeline · Custom
Custom pipeline builder. Compose any sequence of analyst steps with arbitrary context keys and confidence gates.
skill = UniversalSkill.build(
steps=["analyst", "creative", "synthesiser"],
context_keys=["problem"]
)
result = skill.run({"problem": "..."})
