How do I install the Exploration skill?

Run npx skillmds@latest add aretedriver/exploration in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Exploration safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Exploration?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Exploration free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Exploration?

AreteDriver (@aretedriver) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Exploration by aretedriver — Agent Skill

Name: Exploration
Rating: 4.2 (5 reviews)
Author: AreteDriver

Exploration

Exploration-mode prompt scaffold. Use when the option space isn't mapped yet — generate distinct alternatives, surface unknowns, challenge assumptions. The output is a list of options with evidence, not a recommendation. Invoke with /exploration.

AreteDriver Updated Sep 10, 2026

File contents

Exploration Mode

Load the full template from ~/projects/animus/packages/forge/prompts/modes/exploration.md and use it as the structural contract for the output you produce.

Process

Gather INPUTS from the user. If any are missing, ask for them before proceeding. Do not guess. Required slots:
- objective — what the user is trying to decide
- context — known facts, non-negotiable constraints
- assumptions — what the user currently believes (flag for challenge)
- evidence base — docs, data, prior decisions to draw from
- exclusions — already-rejected paths, out-of-scope options
Produce the output in the exact OUTPUT FORMAT specified in exploration.md:
- Option table: id | name | premise | cheapest proof | biggest risk
- Three highest-leverage unknowns (with cheap tests)
- Three stated assumptions that may not hold (with tests)
- One-paragraph synthesis of the option space
Honor the FAILURE CONDITIONS — before emitting, self-check:
- Any two options that are variants of the same idea → collapse or replace
- "It depends" without naming the variable → re-state explicitly
- Recommending a winner → stop, you're in the wrong mode
- No unknowns surfaced → suspicious, re-examine

Anti-pattern

Do not blend this with Evaluation mode. Exploration generates and names; it does not rank or decide. If the user wants a pick, they should invoke /evaluation on the output of this session.

Build-mode goal-loop — disengage

Exploration is the opposite of building. On entry, clear any active goal so the autonomous loop cannot fire during divergent thinking — delete the .goal-active sentinel at the repo root if present:

root=$(git rev-parse --show-toplevel 2>/dev/null) && [ -n "$root" ] && rm -f "$root/.goal-active"

Scoring

If the user wants to score this output against the personal-quality rubric, suggest:

animus-forge eval run <suite> --rubric personal-quality --prompt-version exploration-v1

Weights that matter most for Exploration outputs: precision (specific options, not generic), evidence_quality (options tied to inputs), actionability (each unknown has a cheap test).

---name: explorationdescription: Exploration-mode prompt scaffold. Use when the option space isn't mapped yet — generate distinct alternatives, surface unknowns, challenge assumptions. The output is a list of options with evidence, not a recommendation. Invoke with /exploration.---# Exploration ModeLoad the full template from `~/projects/animus/packages/forge/prompts/modes/exploration.md` and use it as the structural contract for the output you produce.## Process1. **Gather INPUTS from the user.** If any are missing, ask for them before proceeding. Do not guess. Required slots: - `objective` — what the user is trying to decide - `context` — known facts, non-negotiable constraints - `assumptions` — what the user currently believes (flag for challenge) - `evidence base` — docs, data, prior decisions to draw from - `exclusions` — already-rejected paths, out-of-scope options2. **Produce the output in the exact OUTPUT FORMAT** specified in exploration.md: - Option table: id | name | premise | cheapest proof | biggest risk - Three highest-leverage unknowns (with cheap tests) - Three stated assumptions that may not hold (with tests) - One-paragraph synthesis of the option space3. **Honor the FAILURE CONDITIONS** — before emitting, self-check: - Any two options that are variants of the same idea → collapse or replace - "It depends" without naming the variable → re-state explicitly - Recommending a winner → stop, you're in the wrong mode - No unknowns surfaced → suspicious, re-examine## Anti-patternDo not blend this with Evaluation mode. Exploration generates and names; it does not rank or decide. If the user wants a pick, they should invoke `/evaluation` on the output of this session.## Build-mode goal-loop — disengageExploration is the opposite of building. On entry, clear any active goal so the autonomous loop cannot fire during divergent thinking — delete the `.goal-active` sentinel at the repo root if present:```bashroot=$(git rev-parse --show-toplevel 2>/dev/null) && [ -n "$root" ] && rm -f "$root/.goal-active"```## ScoringIf the user wants to score this output against the personal-quality rubric, suggest:```bashanimus-forge eval run <suite> --rubric personal-quality --prompt-version exploration-v1```Weights that matter most for Exploration outputs: `precision` (specific options, not generic), `evidence_quality` (options tied to inputs), `actionability` (each unknown has a cheap test).

Frequently asked questions