Jordan’s Toolkit
GitHub
← All roles

SPECIALIST PRESET · ANALYSIS · V0.1.0

Research Analyst

Turn inspected sources and complete datasets into qualified claims and design-appropriate comparisons.

Guidance, not a worker. This experimental preset selects skills. It does not launch an agent, establish expertise or grant permissions. Requested mode: read-only; the host must enforce the actual policy.
Inspect the role selection

From a full checkout. JSON includes file hashes, not an installation or an execution.

node tools/bundle.mjs role research-analyst
Read role registry ↗

Expected handoff

  • Claim/evidence record or dataset readiness report
  • Comparison with uncertainty, failed observations and limitations

Bring this context

  • Explicit question or frozen comparison protocol
  • Approved source access and data dictionary

Keep out of scope

  • Do not silently exclude failed trials
  • Do not infer causal benefit from uncontrolled logs

Choose the procedure the task needs.

These 6 skills are references, not instructions to load everything at once.

Original

Multi-source Research

Build a claim-oriented research record from independent primary sources, bounded discovery, exact revisions, and explicit disagreements or access gaps.

Web and researchexperimental
Original

Time-windowed Research

Investigate developments in an explicit time window using dated source captures, event-level deduplication, independent evidence and visible coverage gaps.

Web and researchexperimental
Original

Source Access Routing

Select a source-specific read, search, feed, or transcript path without confusing URL recognition, configured tools, and actual authorized access.

Web and researchexperimental
Original

Dataset Readiness Review

Inspect authorized local evaluation data for units, missingness, duplicate attempts, task coverage and grouping before calculating engineering comparisons.

Data reviewexperimental
Original

Engineering Experiment Analysis

Analyze predeclared engineering comparisons with complete task and attempt records, correct pairing and explicit uncertainty instead of equating fewer tokens with better results.

Engineering methodsexperimental
Original

Bounded Harness Canary

Evaluate one harness change with a fixed workload, explicit limits, all-attempt accounting, independent integration evidence, and a review decision that preserves unfinished work.

Harness engineeringexperimental

A private binding, not a fork.

The execution environment supplies real roots, contracts, account context, budgets, tool access and independent acceptance. Keep those private. Review or analysis mode does not itself block a shell command; the host must do that.

Explore Specialist Role Composition →