ahepi/deepreason/example-battery
Claude Code and agent skill by ahepi/deepreason
- community
Checking that the SKILL.md file is still in the repository.
Build a battery of concrete instances BEFORE writing or evaluating any definition, pin, or semantic clause (Reed step 1). Use whenever a term needs a meaning, a pin is being drafted, or an existing pin is being retrofitted. Converts semantic mapping from unaided imagination into curation of juxtaposed structures.
Install
npx vskill@latest install ahepi/deepreason/example-batterySKILL.md
Example Battery (Reed 1)
<!-- PROMPT-CORE-BEGIN --> Before any clause about a term's meaning is written, build its battery.
- Construct, as small concrete structures in the fragment language:
- >=3 POSITIVE instances the intended meaning must admit;
- >=3 NEAR-MISS negatives, each a minimal pair with a positive -
identical except in one respect, which the meaning must exclude;
- >=1 BOUNDARY case whose classification is genuinely undecided,
recorded as OPEN with the question stated.
- Structures are explicit and tiny: name the sorts, list the elements,
give every relevant relation extensionally. No instance is described only in prose.
- Juxtapose everything in one file, positives beside their near-miss
partners, with one line per pair naming the single difference. Near-misses are pairwise DISTINCT in the clause that admits them: no two N-instances may be the same single-bit flip of (structurally) the same positive -- a battery's discriminating power is the count of distinct negative shapes, not the count of N labels. Any structural assumption not present in the inventory candidate (e.g. licensing a chained/transitive reading of a relation) is never a silent positive: it becomes a boundary (B) instance carrying the question, or it is omitted.
- If the term has a recorded risk corridor ("too weak admits X, too
strong excludes Y"), X and Y are MANDATORY battery members: build the structure the too-weak reading wrongly admits and the structure the too-strong reading wrongly excludes. A corridor without its two walls built is not yet analyzed.
- If you cannot construct three positives, the term is not understood
well enough to pin: record what blocks each attempted instance and stop. That record is the deliverable.
- Battery instances carry ids. Digests are produced ONLY by the repo's
digest tool (scripts/battery_digest.py --write) over the canonical instance bytes; NEVER compute, estimate, or invent a digest yourself. Write the literal placeholder PENDING-DIGEST in every digest cell and let the acceptance command run the tool. A hand-written hash is a fabrication even if it happens to be correct.
- The battery file is parsed by machines: its exact grammar (heading
form, id pattern, registry table columns) is defined in zoo/batteries/FORMAT.md and is provided to you as a READ-ONLY REFERENCE section in your context. Conform to it byte-exactly. If that reference section is absent or marked MISSING, STOP and report BLOCKED requesting it -- never invent structure; semantically fine but mechanically unparseable output is a failure.
- Instances are proto-members of the model zoo and feed the denotation
tests. Never delete a battery instance a pin fails on; a failed instance is evidence, not clutter. <!-- PROMPT-CORE-END -->
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/example-battery/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/ahepi/deepreason/example-battery)