https://projectious-work.github.io/processkit/v1.x/docs/skills/Recent content in Skills on processkitHugoenhttps://projectious-work.github.io/processkit/v1.x/docs/skills/format/Mon, 01 Jan 0001 00:00:00 +0000https://projectious-work.github.io/processkit/v1.x/docs/skills/format/<p>This page summarizes the skill package format. The authoritative source is <a href="https://github.com/projectious-work/processkit/blob/main/src/context/skills/FORMAT.md" rel="noopener"><code>src/context/skills/FORMAT.md</code></a> in the processkit repo.</p> <h2 id="directory-layout">Directory layout</h2> <pre tabindex="0"><code>src/context/skills/&lt;category&gt;/&lt;skill-name&gt;/ SKILL.md ← required — three-level agent instructions INDEX.md ← optional — human-readable overview examples/ ← recommended — example outputs templates/ ← recommended — YAML frontmatter entity scaffolds references/ ← optional — deep-dive reference material mcp/ ← optional — Python MCP server server.py mcp-config.json README.md </code></pre><h2 id="required-frontmatter-fields">Required frontmatter fields</h2> <table> <thead> <tr> <th>Field</th> <th>Purpose</th> </tr> </thead> <tbody> <tr> <td><code>apiVersion</code></td> <td>Always <code>processkit.projectious.work/v1</code> at v0.x.</td> </tr> <tr> <td><code>kind</code></td> <td>Always <code>Skill</code>.</td> </tr> <tr> <td><code>metadata.id</code></td> <td><code>SKILL-&lt;skill-name&gt;</code></td> </tr> <tr> <td><code>metadata.name</code></td> <td>Kebab-case; matches directory name</td> </tr> <tr> <td><code>metadata.version</code></td> <td>Semver, independent of processkit release</td> </tr> <tr> <td><code>metadata.created</code></td> <td>ISO 8601 UTC</td> </tr> <tr> <td><code>spec.description</code></td> <td>One-sentence summary (shown in listings)</td> </tr> <tr> <td><code>spec.category</code></td> <td>One of the registered categories</td> </tr> <tr> <td><code>spec.layer</code></td> <td>Integer 0–4, or <code>null</code> for non-process skills</td> </tr> </tbody> </table> <h2 id="optional-fields">Optional fields</h2> <table> <thead> <tr> <th>Field</th> <th>Purpose</th> </tr> </thead> <tbody> <tr> <td><code>spec.uses</code></td> <td>Skills this depends on (strictly lower layer for process skills)</td> </tr> <tr> <td><code>spec.provides</code></td> <td>What the agent gains: primitive kinds, MCP tools, templates</td> </tr> <tr> <td><code>spec.when_to_use</code></td> <td>Trigger description for routing</td> </tr> <tr> <td><code>spec.replaces</code></td> <td>ID of a skill this one overrides (for community forks)</td> </tr> </tbody> </table> <h2 id="categories">Categories</h2> <p><code>process</code>, <code>language</code>, <code>framework</code>, <code>infrastructure</code>, <code>architecture</code>, <code>design</code>, <code>data</code>, <code>ai</code>, <code>api</code>, <code>security</code>, <code>observability</code>, <code>database</code>, <code>performance</code>, <code>meta</code>.</p>https://projectious-work.github.io/processkit/v1.x/docs/skills/hierarchy/Mon, 01 Jan 0001 00:00:00 +0000https://projectious-work.github.io/processkit/v1.x/docs/skills/hierarchy/<p>Process-primitive skills form a strict layered DAG. A skill&rsquo;s <code>spec.uses</code> field may only reference skills in <strong>lower</strong> layers. Cycles are validation errors.</p> <h2 id="the-layers">The layers</h2> <table> <thead> <tr> <th>Layer</th> <th>Role</th> <th>Skills</th> </tr> </thead> <tbody> <tr> <td>0</td> <td>Foundation</td> <td><code>index-management</code>, <code>id-management</code>, <code>event-log</code></td> </tr> <tr> <td>1</td> <td>Primitive management</td> <td><code>role-management</code>, <code>actor-profile</code></td> </tr> <tr> <td>2</td> <td>Core entities</td> <td><code>workitem-management</code>, <code>decision-record</code>, <code>scope-management</code>, <code>category-management</code>, <code>cross-reference-management</code>, <code>binding-management</code></td> </tr> <tr> <td>3</td> <td>Workflow policy</td> <td><code>gate-management</code>, <code>constraint-management</code>; legacy/migration guidance for <code>process-management</code>, <code>state-machine-management</code>, <code>schedule-management</code></td> </tr> <tr> <td>4</td> <td>Cross-cutting</td> <td><code>discussion-management</code>, <code>metrics-management</code></td> </tr> </tbody> </table> <blockquote> <p><strong>Layer 0 has three skills with one intra-layer edge.</strong> <code>index-management</code> and <code>id-management</code> are the absolute foundation — they depend on nothing. <code>event-log</code> is also Layer 0 but <code>uses: [index-management, id-management]</code>, so it conceptually sits &ldquo;atop&rdquo; them. This is the only intra-layer edge in the entire hierarchy. The strict-downward rule applies to Layers 1+ unchanged.</p>