médialab Sciences Po
Official@medialab · Paris, France
SciencesPo's médialab is an interdisciplinary research laboratory gathering engineers, designers & social science researchers.
Agent Skills by médialab Sciences Po
Showing 12 vetted skills indexed across 1 GitHub repositories.
medialabspinosa-writer
Generates numbered markdown research reports from evidence packets and goal artifacts.
medialabagent-interception
Extract and analyze tool-call events from OpenCode and Codex CLI session logs into structured evidence reports.
medialabspinosa-evolver
Applies scoped control-file and behavior-doc updates justified by evaluator audits.
medialabspinosa-evaluator
Audits completed research routes and recommends scoped framework edits.
medialabspinosa-overseer
Audits research workspace coverage and generates gap analysis reports from workspace sources.
medialabspinosa-serendippo
Discovers hidden connections and patterns across raw research files and maps.
medialabspinosa-analyst
Generates contextual analysis reports from project context and prior workflow artifacts.
medialabspinosa-verifier
Verifies claims, quotes, and citations in reports against original source files.
medialabspinosa-janitor
Audits workspace hygiene, detects stale files and broken wikilinks, and proposes archival moves to .trash/.
medialabspinosa-searcher
Searches document corpora with grep and writes structured evidence packets to files.
medialabspinosa-mapper
Extracts content-grounded fragments from raw files and writes Obsidian navigation maps.
medialabsystematic-bugfinder
Diagnose and fix software bugs across five layers from symptom to systemic safeguard.
Frequently Asked Questions About médialab Sciences Po
FAQPage SchemaWhat tasks can I accomplish with médialab Sciences Po's skills?▼
You can run a staged research pipeline: map and extract raw files, search for evidence, analyze context, verify citations against sources, and write final reports. Additional skills audit session logs, clean stale repository files into .trash/, diagnose bugs across five layers, and build Cloudflare Workers applications.
Who should use these skills?▼
Research engineers and social science researchers managing large heterogeneous document corpora benefit most, especially teams needing verifiable, citation-grounded reports. Developers building on Cloudflare's platform—Workers, Pages, D1, R2, Vectorize—also get dedicated coverage, as do maintainers auditing repository hygiene or recurring software defects.
How do the spinosa research skills work together in practice?▼
An orchestrator routes requests through sequential artifact-producing steps: spinosa-mapper extracts fragments, spinosa-searcher writes evidence packets, spinosa-analyst adds context, spinosa-verifier truth-checks claims against sources, and spinosa-writer produces the final numbered report. Evaluator and evolver skills audit and refine the process afterward.
What does the agent-interception skill require and produce?▼
It requires access to OpenCode or Codex session logs. It discovers, parses, and analyzes those logs, producing a structured Markdown evidence report detailing file reads with coverage depth, operation types, search keywords, and call timelines—useful for behavior audits and investigating what happened in a past session.
What Cloudflare capabilities does the cloudflare skill cover?▼
It covers Workers, Pages, KV, D1, R2, Durable Objects, Workers AI, Vectorize, Tunnel, Spectrum, WAF, and DDoS protection, plus Terraform and Pulumi infrastructure definitions. The companion skill adds stateful applications with the Agent class, state management, callable RPC, scheduled tasks, and React hooks.