{{SKILL_ID}}

Purpose

State the reusable capability this skill provides. Define the semantic job, not the implementation mechanism.

Capability Boundary

Use when

• {{OBSERVABLE_TASK_CONDITION_1}}
• {{OBSERVABLE_TASK_CONDITION_2}}

Do not use when

• {{NEGATIVE_SCOPE_CONDITION_1}}
• {{NEGATIVE_SCOPE_CONDITION_2}}

Neighboring artifact classes

State which standards, prompt templates, methodologies, rulesets, personas, state packages, or other skills may be composed with this skill. Do not silently absorb their authority.

Inputs

Required inputs

• {{REQUIRED_INPUT_1}}

Optional inputs

• {{OPTIONAL_INPUT_1}}

Insufficient-input behavior

State what the agent should do when a genuinely required input is absent.

Modes

Delete this section if the skill has only one mode.

{{MODE_A}}

Select when: {{MODE_SELECTION_RULE}}

Describe what changes in this mode.

Supporting Artifacts

List only load-bearing artifacts and explain why they exist.

Artifact	Role	Required?
references/README.md	Domain/reference index	No
scripts/README.md	Helper-script index	No
schemas/README.md	Machine-readable contract index	No
agents/README.md	Runtime-binding notes	No

If a supporting artifact is required for correct execution, explicitly instruct the agent when to read or invoke it.

Procedure

Describe the actual reusable workflow. The procedure may differ completely from other skills.

Step 1 — {{STEP_NAME}}

{{LOAD_BEARING_ACTION}}

Step 2 — {{STEP_NAME}}

{{LOAD_BEARING_ACTION}}

Step 3 — Validate and close

Apply the skill's declared quality gates and verify that the output contract has been satisfied.

Output Contract

The skill must state exactly what it produces.

Primary output

• {{PRIMARY_OUTPUT}}

Additional artifacts

• {{OPTIONAL_OR_REQUIRED_ARTIFACT}}

Output location

State any required output directory or packaging convention. Delete if irrelevant.

Validation and Quality Gates

Completion requires all applicable gates to pass.

• {{QUALITY_GATE_1}}
• {{QUALITY_GATE_2}}
• Output contract is complete.
• No required supporting artifact was skipped.

When correctness is machine-checkable, place the validator in scripts/ or tests/ and reference it here.

Failure and Escalation Behavior

Define what happens when:

• required input is missing;
• a dependency is unavailable;
• validation fails;
• the task falls outside scope;
• another canonical skill is a better fit.

Runtime and Dependencies

Required dependencies

• {{DEPENDENCY_OR_NONE}}

Runtime bindings

Runtime-specific metadata belongs under agents/ or another clearly named adapter layer. Runtime wrappers may adapt presentation but must preserve the semantic capability.

Anti-Patterns

• Do not broaden the skill beyond its declared capability because adjacent work is convenient.
• Do not hide critical procedure in an unreferenced side file.
• Do not treat trigger keywords as stronger than the semantic task.
• Do not let a binary distribution replace inspectable canonical source.
• Add skill-specific anti-patterns here.

Provenance and Lifecycle

Source lineage: {{SOURCE_LINEAGE_OR_NONE}} Supersedes: {{SUPERSEDES_OR_NONE}} Promotion evidence: {{USAGE_AND_VALIDATION_EVIDENCE}}