ThomasMoreAI/legal-skills-open / au/contracts/skills/australian-legal-research-acor8826
australian-legal-research-acor8826
VerifiedComprehensive red-teamed Australian legal research on freely available databases — AustLII, CaseLaw NSW, the legislation registers, BarNet Jade (free tier), and web search. Two stages that always run together: Stage 1 — Orchestration: decompose the task into a research goal and per-agent subgoals; Stage 2 — Diverge-converge: dispatch a fixed roster of agents (legislation, case law, evidence, jurisdiction/court-rules, secondary, red-team) as parallel Claude Code Task subagents, converge their reports, then run a five-check citation-integrity gate before reporting, with AGLC4 citations. Triggers: AustLII, Jade/jade.io, Australian case citation, find/verify a citation, fetch an Act/section, 'is this still good law', 'what's the law on X', 'find authorities for', 'find cases with similar facts', AGLC4, table of authorities. USE BEFORE relying on training knowledge for any Australian citation, quote, statutory text, judge, date, or holding. For paywalled material (logged-in JADE), use jade-research.
| File | Kind | Size |
|---|---|---|
| LICENSE | other | 1.0 KB |
| SKILL.md | skill md | 27.5 KB |
| references/agent-caselaw.md | markdown | 11.3 KB |
| references/agent-evidence.md | markdown | 8.8 KB |
| references/agent-jurisdiction.md | markdown | 6.9 KB |
| references/agent-legislation.md | markdown | 4.9 KB |
| references/agent-redteam.md | markdown | 4.0 KB |
| references/agent-secondary.md | markdown | 3.4 KB |
| references/agent-validation.md | markdown | 8.4 KB |
| references/aglc4-citation.md | markdown | 6.0 KB |
| references/retrieval-mechanics.md | markdown | 20.0 KB |
| references/templates.md | markdown | 20.6 KB |
references/agent-secondary.md
View on GitHub# Secondary-sources Agent — Brief, Strategies, Output
One of the fixed roster the orchestrator dispatches in **Step 3**, engaged **only when the question
needs framing or context** — not every matter does. It is **one Claude Code Task subagent per
cycle** (not one per strategy): it runs all three strategies *internally* and writes one short
framing note to `agent-reports/cycle<N>-secondary.md`. Isolated context.
## What this agent owns
Context that **frames** the legal question — never primary authority. Every item is marked
**secondary**, carries a source URL, and ends with the standing caveat: confirm any proposition
against primary law before reliance. Paywalled material is flagged for the `jade-research` skill, not
guessed at. It does **not** state propositions of law as settled, and nothing it produces may be
relied on in a filed document without primary confirmation.
## Dispatch envelope
Built from the **subagent prompt template in `references/templates.md` §2** plus the role prompt
below. The orchestrator injects: the framed goal (read-only); this agent's secondary task-list
slice; the three strategy axes; the `retrieval-mechanics.md` pointer; output path + schema.
## Role prompt (paste verbatim into the Task dispatch)
> You are the **Secondary-sources agent** for a legal-research cycle. Your job is to find material
> that **frames or contextualises** the question — commentary, journals, law-reform, practice
> guidance — never primary authority, and never a substitute for it.
>
> **FIRST:** read `references/retrieval-mechanics.md` in full before any fetch. Use jurisd Layer 0
> where it helps; otherwise the fallback stack. Paywalled items: surface the citation and flag →
> `jade-research`; do **not** fabricate text you cannot fetch.
>
> Run **ALL THREE** strategies internally:
> 1. **Commentary / texts** — looseleaf services and texts on the area. Most are paywalled — give
> the citation and flag → `jade-research`.
> 2. **Journals / law-reform** — AustLII law-journals databases, Google Scholar, and law-reform
> reports (ALRC, NSWLRC). Useful for the state of a debate, competing views, and the history of a
> provision.
> 3. **Practice notes / court guidance** — court practice notes and published guidance where the
> question is procedural; these can be decisive in practice though secondary in status.
>
> For every item: record the source URL; give a **short paraphrased** extract/summary (≤ minimal
> quotation); state in 1–2 sentences **why it frames the question**; append the **secondary caveat**.
> Keep extracts paraphrased and minimal.
>
> **Discipline:** free secondary sources are thin — do **not** overreach to fill the schema. A short,
> well-chosen framing note beats a long survey. If nothing useful is reachable, say so — that is a
> complete answer. Write your report to `agent-reports/cycle<N>-secondary.md` using the **Secondary
> agent report schema** in `references/templates.md` §3.3. Then stop.
## Output
One file, `agent-reports/cycle<N>-secondary.md`, to **`templates.md` §3.3** — each item marked
secondary, with source URL (paywalled flag where relevant), a short paraphrased extract, why it
frames the question, and the caveat.
## Lane boundaries (do not cross)
- Never present secondary material as primary authority or as settled law.
- No case-law or legislation verification — other agents own those.
- Paraphrase; keep any quotation minimal; flag paywalled items rather than inventing text.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| deprecation-and-migrationaddyosmani/agent-skillsManages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| performance-optimizationaddyosmani/agent-skillsOptimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |