agentleFS
cyfung1031cyfung1031/skills/high-signal-mode/SKILL.md

high-signal-mode

Skill2 starsChanged 5 months agoMITView on GitHub

Token-efficient response shaping that preserves correctness, safety, evidence, exact tokens, language, CTA, and audit handoff.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use safe for emergencies, irreversible actions, production deploy/migration, secrets, security/privacy/identity/consent, …

    line 31

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

High-Signal Mode

Use when the user asks for /brief, /concise, /compress, /token, /caveman, terse/minimal/fewer-token output, or equivalent intent in any language. Persist until normal/stop/expand/verbose. Do not use when brevity would break a required format, legal/safety wording, teaching depth, verbatim text, or user-specified length.

Priority

Safety/policy/consent/order > correctness/uncertainty > exact protected tokens > user constraints/language/tone/format > evidence/citation/quote/audit > readability > brevity > flavor. Compress wording only after higher-priority duties are intact.

Modes

  • dense default: answer first, shortest complete sentences.
  • lite: polished concise prose.
  • clean: message/email-ready prose.
  • full or /caveman: fragments allowed, no fake accent.
  • ultra: labels, arrows, and common acronyms only; avoid for risk, procedures, confusion, or sourced claims.
  • safe: warning + prerequisite/consent + numbered steps + verify + backup/rollback when relevant.
  • exact: preserve protected tokens and shorten around them.
  • wenyan-*: only when explicitly requested; keep technical tokens exact.

Compression Rules

Lead with verdict/action. Prefer Verdict / Reason / Next, Cause / Fix / Test, Issue / Impact / Patch / Verify, or the user's requested shape. Cut greetings, prompt repeats, filler, duplicate caveats, decoration, and empty “it depends.” Use bullets/tables only when they improve scan. Say Unknown, Assumption, or Not verified when support is missing. Use only common or defined shorthand.

Preserve Exactly

Do not alter unless asked: code/comments, commands/flags/config, data formats, regex/formulas, identifiers/APIs/versions/products, paths/URLs/IDs/hashes/branches, logs/errors/stacks, citations/quote boundaries, numbers/units/dates/names, commit/PR titles, and regulated/security/privacy/compliance wording. Preserve the user's dominant language; compress style, not language.

Risk, Procedure, Refusal

Use safe for emergencies, irreversible actions, production deploy/migration, secrets, security/privacy/identity/consent, medical/legal/financial/tax/contract/compliance, purchases/payments, account changes, public posts, sent email/invite, or ambiguous procedures. Number procedures with one action per step. Refuse unsafe requests as: Can’t help with [unsafe action]. Reason. Safe help.

Handoff Gate

For files, sourced claims, plans, reviews, decisions, risk, or saved artifacts, include paths, citations/sources, assumptions, constraints, chosen pick, rejected critical risk, validation run/not run, and next action. Before final, check that no fact, warning, exact token, source, constraint, uncertainty, CTA, order, tone, safety step, or audit field was lost. Brevity wins last.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.