Index of every page on this site: llms.txt
All skills
cli avatar

/cli-exercise-exact

@f24a94f
by GitHub CLIcli/cli47k stars
9,433
Run it yourself

Internal exact-case flow for cli-exercise. Preserves supplied commands, inputs, ordering, assertions, and output constraints.

Use this Skill: https://skilld.dev/gh/cli/cli/cli-exercise-exact

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈36 tokens for metadata: the name and description. ≈774 when used: this file.

Description uses 1.8% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈146 characters, or 1.8%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Record exact cases

Require the entry skill's preserved request, validated contracts, ready preflight receipt, and approved effect scope. If any is missing, return to the entry skill instead of filling gaps with assumptions.

Normalize without changing meaning

Keep the caller's original case text. Turn supplied commands and interactions into the canonical actions in the contract. Keep test ordering and case IDs stable.

Each recording case covers one behavior, with at most one exercise invocation. Give it a very short behavior/outcome title, not an inventory of commands or flags. A supplied case may include multiple preparation or validation commands. If its exact steps conflict with this structure, ask before splitting, reordering, omitting, or adding anything. Explicit narrower coverage takes precedence over the default preparation/exercise/validation coverage.

If the caller specifies literal keys, preserve those keys. If they specify an option by name, a mechanical selector can locate it. Ask before inventing values or steps that materially affect the case. Do not add negative tests, extra metadata, new retries, or exploratory branches just because they seem useful.

Keep before/after conditions controlled: equivalent initial state, commands, geometry, data, and timing, with separately identified executables.

Execute faithfully

Use the shared run module. The exact-action gate must remain active for both live AI actions and generated controllers.

Execute each case's preparation, exercise, and validation in that order before starting the next case; approved cleanup can follow its validation. Do not batch phases across cases or later rearrange captured runs to suggest a different execution order. Preserve and clarify conflicting caller-supplied ordering.

Observe the real screen before actions. Condition-based waits can synchronize unspecified timing, but cannot replace or defeat a stated timing requirement. Never match an expected output merely because it appears in the typed command.

An unexpected prompt is not permission to reinterpret the test. Use only mechanical actions consistent with the supplied case. Otherwise pause, preserve evidence, and report what needs clarification.

If an authorized write is part of the case, execute it as specified or report blocked. Do not replace it with Cancel and claim that the case passed. Follow all applicable environment/tool policies.

Judge the specified assertions

Check every expectation against actual observations, with references to the appropriate state, output, file, or approved external readback. Do not rewrite the expectation to match the observed result.

Use failed for contradicted expectations and blocked for requirements that could not be exercised or verified. An execution with no correctness assertion is observed, not passed. A nonzero exit can pass an expected-error case; exit zero cannot excuse a contradicted assertion.

Finish with evidence and cleanup.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warningtoday3 checks · Risk SAFE
  • Gen Agent Trust Hubtoday

    The skill defines an internal workflow for executing and recording CLI testing cases. It focuses on maintaining the fidelity of supplied commands and assertions and contains no malicious content.

  • Sockettoday

    No alerts

  • Snyktoday

    Risk: MEDIUM · 1 issue

Signed by skilld at f24a94f. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 2 months ago
user-invocable
false

README badge

README badge for cli/cli/cli-exercise-exact