Read and present NeuroTech intelligence briefings when the user asks about neurotech, BCI news, or recent reports.
This skill gives you access to the output of the NeuroTech NewsHound pipeline, which runs nightly and produces intelligence briefings on implantable brain-computer interfaces, ECoG/sEEG, microstimulation, and enabling technologies.
Activate when the user asks about:
All files live in this skill's directory (neurotech_reports/).
Each run produces four files with the naming pattern YYYY-MM-DD.*:
| Pattern | Contents |
|---|---|
YYYY-MM-DD.md | Full Markdown report: executive brief, themes, scored items, reviewer notes |
YYYY-MM-DD.html | Styled HTML version of the same report |
YYYY-MM-DD.alerts.json | JSON array of high-priority items (scored 9-10) |
YYYY-MM-DD.full.json | Complete pipeline output: all scored items, themes, errors, metadata |
To find the latest report: sort the .md files by date in the filename and read the most recent one.
| File | Contents |
|---|---|
dashboard.html | Cross-run summary dashboard |
discoveries.yaml | New companies/entities discovered by the pipeline, pending human review |
meta_actions.yaml | Trace of the meta-agent's self-improvement actions (vocabulary updates, source health checks, coverage assessments) |
cron.log | Raw log of nightly pipeline runs |
| File | Contents |
|---|---|
config.yaml | Pipeline configuration: sources, watchlist, models, scoring parameters |
vocabulary.yaml | Domain vocabulary (130+ terms) used for PubMed queries and regex filtering |
discovery_memory.json | Persistent entity tracker: companies, labs, technologies seen across runs |
editorial_memory.json | Story-level memory: what has been reported before (for dedup and novelty classification) |
.md file by date.md or .full.jsondiscoveries.yaml and discovery_memory.json for tracked entities.alerts.jsonmeta_actions.yaml for the latest self-improvement tracecron.log tail for run health (duration, token cost, item counts)