Prompt

Armor for coding agents: a playbook of Claude Code skills that make an agent prove its work, not describe it.

Install

npx -y skills add yiyaw-lab/agent-armor --skill prompt

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Personal prompt library + prompt compiler. Save frequently used prompts as parameterized templates, invoke them by name, or refine a draft prompt into a more effective, token-lean version — informed by TASTE.md so the refined prompt asks for what the user will actually accept, not just what they typed. Modes: <name> | save | list | refine.

SKILL.md

2.9 KB, as published. Nobody here has run it

The library lives in ~/.claude/prompts/, one file per prompt: frontmatter (name, when — one line on when to reach for it) + body = the template, with {placeholders} for the parts that vary.

/prompt <name> [args]

Read ~/.claude/prompts/<name>.md, fill {placeholders} from args and session context, then EXECUTE the template as if the user had typed it in full. Infer missing placeholders from context; ask only if genuinely ambiguous. (This is the token saver: the user types 3 words, the agent runs the 80-word version that actually works.)

/prompt save [name] [text]

Source: the given text, or the user's most recent substantive prompt this session.
Distill to a reusable template: replace session-specific details with {placeholders}; KEEP the phrasing that made it effective (constraints, the bar, the verification ask).
Write the file, report: name, when-to-use, and an invocation example.

/prompt list

Table: name | when to use | example invocation. If the library is empty, say so and point at save/refine.

/prompt refine <draft | "last">

Compile the prompt for effect-per-token. Diagnose, then rebuild:

Goal, not artifact. State the outcome wanted, not just the deliverable. (The user's revealed pattern: the real bar arrives as round-2 corrections — front-load it instead.)
Acceptance criteria from TASTE.md. Read ~/.claude/TASTE.md ACTIVE rules and bake the relevant ones into the prompt explicitly — e.g. modern design idiom, verified-not-claimed side effects, system-level scope, loose ends closed. This is the difference between what the user types and what they'll accept.
Cut filler. Pleasantries, hedges, context the agent already has, double-asks.
Structure. Constraints and output format as terse bullets; one primary ask per prompt.

Output: BEFORE (word count) → AFTER (word count), then a 2–3 bullet "what changed and why."
Rule: the refined prompt must not be longer than the original unless missing acceptance criteria justify the growth — effectiveness first, tokens second. Never pad.
End by offering /prompt save <suggested-name> if the shape looks reusable.

Rules

Templates are the user's voice, not corporate prompt-speak — terse, direct, no "please act as."
Never execute a refine result without the user sending it; refine produces text, not action.
The library is local. Publishing it is a disclosure decision (prompts can leak project strategy).