hegl

Plugin hegl from marketplace hegl in Katch-AI/hegl-plugin. Commands read from the repo's manifest — yesterday.

Skills (14)

• activepieces

  Use when building or editing any Activepieces flow — trigger flows, the candidate flow, piece steps, CODE steps, sandbox steps, routers, loops — when installing or referencing a piece, when sampling a piece action's output, when reading flow JSON shapes or calling the `activepieces_*` tools, or when

• asciidoc

  Use when writing an artifact body (the AsciiDoc documents recorded with the Hegl piece's `record_artifact` action and validated with `validate_artifact`), or when a rendered artifact looks wrong in the dashboard.

• building-the-automation

  Use when asked to build the user's automation and no candidate flow exists yet. Covers learning the job, the design decisions a flow needs, the plan documents, the build order, and the smoke test. Hands off to implementing-decision-boxes.

• calling-custom-apis

  Use when the customer's agent must call an HTTP API that has no dedicated Activepieces piece — internal services, undocumented vendor APIs, anything reachable over plain HTTP(S) with API-key, basic, OAuth2, or JWT-from-shared-secret auth. NOT for APIs that have a real piece (search the pieces catalo

• composing-email-html

  Use when authoring the htmlBody of any send_email step — during Execute when building flow steps that send email, and during Plan when sketching email content. Covers the fragment contract (Hegl wraps every email in a branded layout), what styling comes free, the palette, and copy-paste snippet patt

• feedback

  Use when the user runs `/hegl:feedback` or asks to act on the feedback an agent has received.

• fetching-data-via-pieces

  Use when you need records from an external system the organization has a connection for — pulling past examples for an eval set, checking what a connection returns, any ad-hoc read through a piece action (Salesforce, Gmail, JIRA, HubSpot…).

• fix-errors

  Use when the user runs `/hegl:fix-errors` or asks why an agent's runs are failing, or about an incident.

• hegl

  Use when starting any work on a Hegl agent (building an automation, improving a decision box, inspecting runs or feedback) and whenever unsure which Hegl skill or tool applies. Orientation and the rules every other Hegl skill assumes.

• implementing-decision-boxes

  Use when the candidate flow is scaffolded and smoke-tested, to turn stub decision boxes into real implementations and improve them. Covers eval sets, the experiment backlog, iterations, metrics, porting the winner into the flow, and finishing a version for the user's review.

• linking-to-the-dashboard

  Use when you mention anything the user can look at in the Hegl dashboard (an agent, a run, a version, a decision box, an experiment, a form, a connection, usage, billing) or hand them a step only they can take. Covers where each link comes from, when to open it, and request links that expire.

• mermaid

  Use when writing or editing any Mermaid diagram the user will see — the flow diagram in plan.md, anything pushed to flowDescription, a diagram in a spec.

• platform-api

  Use when deciding which Hegl tool performs an operation (agent documents, decision boxes, flows, trigger flows, connections and config, piece installs, improvement runs, experiments, eval datasets, iterations, metrics, feedback, email, sandboxes, artifacts, platform issues), when a step has no tool

• rendering-templates

  Use when a flow must compose human-readable text from data — an email body, a Slack message, a Markdown summary, a plain-text notice. NOT for building JSON or API payloads (pass objects through step outputs or a CODE step instead).

MCP servers (1)

• hegl

  connects to: https://app.hegl.ai/mcp

Hooks (2)

• PostToolUse

  node "${CLAUDE_PLUGIN_ROOT}/hooks/open-dashboard.mjs"

• PostToolUseFailure

  node "${CLAUDE_PLUGIN_ROOT}/hooks/open-dashboard.mjs"

Similar plugins