salesforce/zana/salesforce-toolkit

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Use the optional sf-agentic-tools provider for Salesforce, Agent Script, Flow, Code Analyzer, metadata and Data 360 operations through Zana's scoped gateway.

Install

npx vskill@latest install salesforce/zana/salesforce-toolkit

SKILL.md

Salesforce toolkit

Use sf_tools to discover and execute the pinned sf-agentic-tools SDK:

  1. { "action": "list" } lists the complete catalog and current provider mode.
  2. { "action": "describe", "tool": "sf_flow" } returns its input schema, action availability and prerequisites.
  3. { "action": "call", "tool": "sf_flow", "input": { "action": "quality.rules" } } executes an action.
  4. { "action": "result.read", "runId": "<returned ID>", "pointer": "/data", "offset": 0, "limit": 20 } retrieves retained evidence.

sf_flow, code_analyzer and sf_metadata also accept toolkit inputs directly. Use the gateway for toolkit sf_soql, sf_apex, sf_lwc, all four agentscript_* tools and eleven data360_* tools. In Both mode the original sf_soql, sf_apex, sf_lwc and sf_agent contracts remain available directly. Schemas differ: always describe the toolkit action before calling it.

The host supplies the registered local project and selected org. Use sf_workbench context.select to change the target. Do not supply workspace, target_org, credentials, artifactDir, resume paths or allowEffects. File inputs and outputs must stay inside the project. Toolkit execution on a remote project requires a future host adapter; discovery remains available.

Read diagnostics, artifacts and nextActions before retrying. Follow prerequisite failures with the specific setup action. Do not claim an unavailable capability worked. Effectful operations (tests, fixes, exports, publish, activation and live preview) require the host approval interaction, even on sandbox orgs. Never retry a refused operation with alternate authorization flags.

Results are bounded. Keep the returned runId to read evidence or provide it as resumeId on a subsequent call (or a direct toolkit tool). IDs are valid only for this plugin activation and original project/org. Up to 32 runs are retained.

CLI equivalent:

sh
zcc sf tool sf_tools --input '{"action":"describe","tool":"sf_flow"}' --json
zcc sf tool sf_flow --input '{"action":"quality.rules"}' --json

Plugins → Salesforce → Agent tool provider selects Both (default), Built-in or Toolkit. sf_workbench remains available in every mode. Built-in mode loads no toolkit runtime and refuses stale toolkit calls; Toolkit mode refuses stale calls to the original family tools. New agent catalogs reflect the selection.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
38
Repository forks
10
7-day trend score
0

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/salesforce/zana/salesforce-toolkit/badge)](https://verified-skill.com/skills/salesforce/zana/salesforce-toolkit)

Something wrong with this skill or its page? Tell us.