agentleFS
jammyfujammyfu/open-game-skills/skills/disciplines/action-feel/SKILL.md

action-feel

Skill13 starsChanged 24 days agoMITView on GitHub

Use when attacks feel floaty, cancels fire illegally, hitstop freezes unrelated actors, or combat outcomes change with render rate or same-tick processing order.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not combine columns implicitly.

    line 17
  2. Distinctive:

    update clocks that are not frozen and resolve legal state/cancel transitions.

    line 26
  3. Distinctive:

    Every contact carries a stable attack/hit instance identity.

    line 34

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Action feel

Choose the project column before tuning numbers. Studio labels are examples, not rules.

ColumnBufferCancelHitstopTurn
short-specialshort attack bufferon-hit into named specialsauthored per contactsnap on startup if published
long-cancellonger command buffernamed on-hit/jump/ranged edgesauthored rhythm/weightoptional inertia carry
commit-whitelistanti-misinput buffernamed follow-ups onlyauthored weightlimited during commitment
coyote-platformerjump buffer + coyoteonly published attack/jump edgestiny or nonepublished air control

Do not combine columns implicitly. Persist the selected values in project data.

Logical clock and phase order

Publish the logical tick rate; render FPS is never the combat clock. Engine adapters provide input, pose, contact-query and presentation hooks, but gameplay owns ordering.

A deterministic default phase is:

  1. sample input into the input ring;
  2. update clocks that are not frozen and resolve legal state/cancel transitions;
  3. snapshot active hit/grab volumes and collect contact candidates;
  4. resolve same-tick contacts using the published trade/priority rule and stable IDs;
  5. apply damage, reaction, hitstop and queued launch state;
  6. emit presentation events after logical results are known.

A project may choose another order, but it must publish it and regression-test same-tick cases.

Every contact carries a stable attack/hit instance identity. A target is hit at most once by one hit instance unless the move explicitly schedules separate multi-hit indices. Container iteration order must not decide who wins a trade.

Hitstop freezes only the actor clocks named by the resolved event. Input sampling, unrelated actors and the world keep running unless a different pause system explicitly owns them. Projectiles or hazards without an actor clock are not accidentally frozen because two actors touched.

Charge and cancel ownership

Publish charge start → full → hold-cost → release → cancel. Interruption, weapon swap and focus loss must restore a legal state.

Cancels are directed edges plus windows and conditions. Motion commands are stepwise sequences with explicit timing windows, not an unordered bag of recent inputs.

Accept

Replay the same input/contact trace at different render rates and with reversed entity iteration order. Move starts, legal cancels, contact IDs, trades, hitstop ownership and resulting states must match. Freeze A/B and verify unrelated C still advances. Record which scenarios actually ran; a timing table alone is not runtime evidence.

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.