Skip to content

PranavNagrecha/AwesomeSalesforceSkills / .claude/skills/salesforce-agentforce

salesforce-agentforce

Verified

Router for the 53 SfSkills `agentforce` skill packages. Agentforce and Einstein: agents, topics, actions, prompt templates, grounding, guardrails, evaluation and production readiness. Use when the request mentions Agentforce, agent topic, agent action, prompt builder, Einstein, Trust Layer, grounding, RAG, guardrails, agent evaluation, prompt injection. Finds and opens the exact skill package to read; it does not contain the guidance itself.

FileKindSize
SKILL.mdskill md5.1 KB
references/skill-index.mdmarkdown15.4 KB
---
name: salesforce-agentforce
description: "Router for the 53 SfSkills `agentforce` skill packages. Agentforce and Einstein: agents, topics, actions, prompt templates, grounding, guardrails, evaluation and production readiness. Use when the request mentions Agentforce, agent topic, agent action, prompt builder, Einstein, Trust Layer, grounding, RAG, guardrails, agent evaluation, prompt injection. Finds and opens the exact skill package to read; it does not contain the guidance itself."
---

# Salesforce agentforce — SfSkills domain router

Agentforce and Einstein: agents, topics, actions, prompt templates, grounding, guardrails, evaluation and production readiness.

**53 skill packages** live under
`${CLAUDE_PLUGIN_ROOT}/skills/agentforce/<slug>/SKILL.md`. They are not
loaded — reach them by path, on demand.

**Generated by `scripts/build_plugin.py`. Do not hand-edit.**

## How to find the right skill

Three mechanisms, listed in order of reliability on a fresh install.
Use the first one that is available; do not stop at a guess.

**1. The shipped roster (always works, no setup).**
Read `references/skill-index.md` next to this file. It lists every
`agentforce` skill package with a one-line gloss, generated from
`registry/skills.json`. Scan it and pick by name.

**2. The MCP server (fast, needs the `sfskills-mcp` server connected).**
Call the `search_skill` tool with the user's phrasing and `domain: "agentforce"`. It returns
ranked skill ids. `get_skill` then returns the package contents.

**3. The search CLI (fast, needs a locally built index).**

```bash
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/search_knowledge.py" "<the user's question>" --domain agentforce --json
```

This needs `vector_index/`, which is **not shipped** — it is gitignored
and must be built once per clone:

```bash
cd "${CLAUDE_PLUGIN_ROOT}" && python3 -m pip install -r requirements.txt && python3 scripts/build_index.py
```

If the command errors or reports `Coverage: NONE`, fall back to
mechanism 1 rather than telling the user the topic is uncovered.

**Then read the package.** Open the exact
`${CLAUDE_PLUGIN_ROOT}/skills/<domain>/<slug>/SKILL.md` the lookup
returned, plus its `references/gotchas.md` and
`references/llm-anti-patterns.md`. Do not answer from this router:
it is a map, not the territory.

## Featured entry points

Curated starting points when the request is broad or the lookup is
ambiguous. This is a shortlist, not the catalogue — the roster at
`references/skill-index.md` has all 53.

- `${CLAUDE_PLUGIN_ROOT}/skills/agentforce/agentforce-agent-creation/SKILL.md` — the end-to-end shape of an Agentforce agent before you write an action
- `${CLAUDE_PLUGIN_ROOT}/skills/agentforce/agent-topic-design/SKILL.md` — topic scoping and instructions — the single biggest driver of agent quality
- `${CLAUDE_PLUGIN_ROOT}/skills/agentforce/agent-actions/SKILL.md` — action design, slot filling, and what belongs in Apex vs Flow
- `${CLAUDE_PLUGIN_ROOT}/skills/agentforce/agentforce-guardrails/SKILL.md` — keeping an agent inside its lane, with refusal behaviour
- `${CLAUDE_PLUGIN_ROOT}/skills/agentforce/prompt-builder-templates/SKILL.md` — prompt templates, grounding fields, and versioning them safely
- `${CLAUDE_PLUGIN_ROOT}/skills/agentforce/einstein-trust-layer/SKILL.md` — masking, zero-retention, and the audit trail your security team will ask for
- `${CLAUDE_PLUGIN_ROOT}/skills/agentforce/agentforce-testing-strategy/SKILL.md` — how to test a non-deterministic agent without hand-checking transcripts
- `${CLAUDE_PLUGIN_ROOT}/skills/agentforce/agentforce-production-readiness-checklist/SKILL.md` — the pre-launch gate for anything customer-facing

## Decision trees

Read the tree *before* activating a skill when the request could be
solved more than one way, and cite the branch that decided it.

- `${CLAUDE_PLUGIN_ROOT}/standards/decision-trees/agentforce-capability-selector.md` — Agentforce vs Prompt Builder vs Next Best Action vs Model Builder
- `${CLAUDE_PLUGIN_ROOT}/standards/decision-trees/automation-selection.md` — read when an agent action could equally be a Flow or Apex

## Canonical templates

Do not hand-roll an idiom that already exists. Copy from
`${CLAUDE_PLUGIN_ROOT}/templates/agentforce/` and rename in the consuming
project; never edit the template in place.

- `${CLAUDE_PLUGIN_ROOT}/templates/agentforce/AgentActionSkeleton.cls`
- `${CLAUDE_PLUGIN_ROOT}/templates/agentforce/AgentTopic_Template.md`
- `${CLAUDE_PLUGIN_ROOT}/templates/agentforce/AgentEval_Fixture.md`
- `${CLAUDE_PLUGIN_ROOT}/templates/agentforce/README.md` — the full list

## Run-time agents for this domain

Invoke one of these subagents when the ask is a whole workflow
rather than a single question:

- `agentforce-action-reviewer` — Review Agentforce agent/action quality
- `agentforce-builder` — Scaffold a complete Agentforce action
- `agentforce-quality-engineer` — Agentforce Quality Engineer

## Rules

1. Answer from the opened `agentforce` package, never from this router.
2. Cite the skill id and, where one applied, the decision-tree branch.
3. Never claim a topic is uncovered without pasting lookup output.
4. Never deploy to an org.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 323kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 323kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 323kUpdated Claude CodeCodexVerifiedInstall