docs(skills): default routine Astra work to medium reasoning

[BramVR]

Routine Astra work now defaults to medium reasoning across PStack skills and Poteto playbooks. High reasoning remains assigned to reviews, judges, and demanding work such as architectural or algorithmic tradeoffs, ambiguous root-cause analysis, and complex concurrency or performance diagnosis. Task size or duration alone does not justify high.

Updates the shared model contract, Poteto subagent instructions, implementation playbooks, How/Why synthesis, Arena candidates, and verification lanes so caller instructions preserve the selected effort. Autoreview remains Astra high with Sol fallback. Existing Sol, Terra, and Luna assignments and the no-fast-mode rule remain unchanged.

Validation: all 89 skills passed YAML/frontmatter validation; git diff --check passed. Manually checked role and caller consistency, verified model pools were preserved, and confirmed 42 installed file paths match across Codex, Agents, and Claude mirrors. Changes are Markdown instructions only.

[Copilot]

🟢 Approval recommended

Documentation-only updates with one minor typo fix suggested; no behavioral/runtime risk introduced.

Pull request overview

Updates skill/playbook model contracts so routine gpt-6-astra work defaults to medium reasoning_effort, while keeping high reserved for reviews and explicitly reasoning-heavy work (tradeoffs, ambiguous root cause, concurrency/perf diagnosis), without changing the “no fast mode” constraint.

Changes:

• Adjusted how / why contracts to require explicit reasoning-effort selection by role, with Astra defaulting to medium outside heavyweight cases.
• Updated Poteto playbooks + poteto-agent reference to encode Astra medium-by-default routing and when to elevate to high.
• Updated arena / figure-it-out guidance and recorded the policy shift in CHANGELOG.md.

File summaries

File	Description
skills/why/SKILL.md	Updates why agent contract: Astra synthesizer defaults to medium; Luna investigators remain high.
skills/how/SKILL.md	Updates how agent contract: Astra explain/synthesize defaults to medium; critics remain high.
skills/poteto-mode/SKILL.md	Updates Poteto “Model routing” to set Astra medium-by-default with explicit high-only criteria.
skills/poteto-mode/references/poteto-agent.md	Makes subagent instructions honor parent effort; Astra defaults to medium unless elevated.
skills/poteto-mode/playbooks/refactoring.md	Updates delegation guidance to Astra medium reasoning.
skills/poteto-mode/playbooks/perf-issue.md	Defaults perf delegation to Astra medium, with explicit high exception for complex perf/concurrency reasoning.
skills/poteto-mode/playbooks/orchestrate.md	Updates worker/verifier routing to reflect Astra medium for routine work, high for judgment-heavy verification.
skills/poteto-mode/playbooks/multi-phase-plan.md	Updates swarm lane effort selection (Astra medium for gates/perf unless diagnosis warrants high).
skills/poteto-mode/playbooks/hillclimb.md	Updates delegation guidance to Astra medium with explicit “raise to high” criteria.
skills/poteto-mode/playbooks/feature.md	Updates code-writing delegation guidance to Astra medium with “raise to high” criteria.
skills/poteto-mode/playbooks/eval.md	Aligns candidate spawn instructions to follow Arena’s declared model+effort contracts.
skills/poteto-mode/playbooks/bug-fix.md	Updates fix delegation guidance to Astra medium with explicit “raise to high” criteria.
skills/figure-it-out/SKILL.md	Updates architect/arena guidance to follow Arena contracts; calls out Astra high for one-way-door design decisions.
skills/arena/SKILL.md	Updates runner effort policy (Astra medium default, high for review/heavy candidates) and Phase B wording about explicit effort.
CHANGELOG.md	Adds 2026-09-06 entry documenting the Astra reasoning default change.

Review details

• Files reviewed: 15/15 changed files
• Comments generated: 1
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.