agentforce-generate

Name: agentforce-generate
Author: SalesforceAIResearch

180installs37forks

How to install agentforce-generate?

salesforceairesearch/agentforce-adlc/agentforce-generate

$npx -y skills add SalesforceAIResearch/agentforce-adlc --skill agentforce-generate

Installs into the current project.

Summary

Build, modify, optimize, debug, and deploy agents with Agentforce Agent Script. TRIGGER when: user creates, modifies, optimizes, or asks about .agent files or aiAuthoringBundle metadata; changes agent behavior, responses, or conversation logic; designs agent actions, tools, subag

Files · 1

View on GitHub

SKILL.md

1	# Agent Script Skill
2
3	## What This Skill Is For
4
5	This skill is for developing Agentforce agents, primarily with Agent Script, Salesforce's scripting language for AI agents.
6
7	CRITICAL: Agent Script is NOT AppleScript, JavaScript, Python, or any other
8	language. Do NOT confuse Agent Script syntax or semantics with any other
9	language you have been trained on.
10
11	Agent Script agents are defined by `AiAuthoringBundle` metadata: a `.agent` file (agent behavior) plus `bundle-meta.xml` (bundle metadata). Actions can be implemented with invocable Apex, autolaunched Flows, Prompt Templates, and other supported types.
12
13	This skill covers the full Agent Script lifecycle: designing agents,
14	writing Agent Script code, validating and debugging, deploying and
15	publishing, and testing.
16
17	## How to Use This Skill
18
19	This file maps user intent to task domains and relevant reference files in `references/`. Treat this file as the execution router for end-to-end agent development, and use references for deep detail.
20
21	Identify user intent from task descriptions. ALWAYS read indicated reference files BEFORE starting work.
22
23	## Rules That Always Apply
24
25	1. Always `--json`. ALWAYS include `--json` on EVERY `sf` CLI command. Do NOT pipe CLI output through `jq` or `2>/dev/null`. Read the full JSON response directly — LLMs parse JSON natively.
26
27	2. Verify target org. Before any org interaction, run `sf config get target-org --json` to confirm a target org is set. If none configured, ask the user to set one with `sf config set target-org <alias>`.
28
29	3. Diagnose before you fix. When validating/debugging agent behavior,
30	ALWAYS `--use-live-actions` to preview authoring bundles. Send utterances
31	then read resulting session traces to ground your understanding of the
32	agent's behavior. Trace files reveal subagent selection, action I/O, and
33	LLM reasoning. DO NOT modify `.agent` files or action implementations without
34	this grounding. See [Validation & Debugging](references/agent-validation-and-debugging.md)
35	for trace file locations and diagnostic patterns.
36
37	4. Spec approval is a hard gate. Never proceed past Agent Spec
38	creation without explicit user approval.
39
40	5. Don't stall. After a step completes successfully, announce the
41	next step and start it. Do not wait for the user to say "what's next"
42	or "ok, continue." The only checkpoints that require explicit user
43	approval are: (a) Agent Spec approval, (b) the pre-Publish CHECKPOINT,
44	(c) any A/B branch the skill explicitly surfaces (e.g., Data Cloud
45	not provisioned during ADL setup). Long-running async work like ADL
46	indexing should run in the background while the skill continues with
47	work that doesn't depend on the result.
48
49	6. Draft-first lifecycle. During normal authoring, stay in draft iteration:
50	edit `.agent` + action implementations, validate, deploy, and preview as many
51	times as needed. Do NOT publish/activate by default. Publish + activate are
52	explicit release actions that require the user to confirm they are ready to
53	commit the current draft to metadata and expose it to end users.
54
55	7. Default agentic, pin with cause. Use the most agentic posture that meets
56	each subagent's requirement, and add deterministic controls only for
57	regulation/trust gates or observed failures. For detailed posture rules, see
58	[Posture & Determinism](references/posture-and-determinism.md).
59
60	8. No nested `if` or `else if`. Agent Script only supports flat `if`/`else` blocks. No `else if`, no `if` inside `else`, no `if` inside `if`. For multi-branch logic, use sequential `if` statements or compound conditions (`if A and B:`). Nested structures cause silent compile failures.
61
62	9. Action implementation is a user decision. During planning/spec work,
63	default new actions to `NEEDS STUB` placeholders. Always ask the user whether
64	they want to scan org/project for existing implementations and/or generate
65	new Apex/Flow/Prompt implementations before taking either path.
66
67	## Task Domains
68
69	Every task domain below has Required Steps. Follow verbatim, in order. The default path is: design -> draft implementation lo

People who liked this also viewed

6 picks

Type

agentforce-generate

180installs37forks

View on GitHub

How to install agentforce-generate?

salesforceairesearch/agentforce-adlc/agentforce-generate

$npx -y skills add SalesforceAIResearch/agentforce-adlc --skill agentforce-generate

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Agent Script Skill
2
3	## What This Skill Is For
4
5	This skill is for developing Agentforce agents, primarily with Agent Script, Salesforce's scripting language for AI agents.
6
7	CRITICAL: Agent Script is NOT AppleScript, JavaScript, Python, or any other
8	language. Do NOT confuse Agent Script syntax or semantics with any other
9	language you have been trained on.
10
11	Agent Script agents are defined by `AiAuthoringBundle` metadata: a `.agent` file (agent behavior) plus `bundle-meta.xml` (bundle metadata). Actions can be implemented with invocable Apex, autolaunched Flows, Prompt Templates, and other supported types.
12
13	This skill covers the full Agent Script lifecycle: designing agents,
14	writing Agent Script code, validating and debugging, deploying and
15	publishing, and testing.
16
17	## How to Use This Skill
18
19	This file maps user intent to task domains and relevant reference files in `references/`. Treat this file as the execution router for end-to-end agent development, and use references for deep detail.
20
21	Identify user intent from task descriptions. ALWAYS read indicated reference files BEFORE starting work.
22
23	## Rules That Always Apply
24
25	1. Always `--json`. ALWAYS include `--json` on EVERY `sf` CLI command. Do NOT pipe CLI output through `jq` or `2>/dev/null`. Read the full JSON response directly — LLMs parse JSON natively.
26
27	2. Verify target org. Before any org interaction, run `sf config get target-org --json` to confirm a target org is set. If none configured, ask the user to set one with `sf config set target-org <alias>`.
28
29	3. Diagnose before you fix. When validating/debugging agent behavior,
30	ALWAYS `--use-live-actions` to preview authoring bundles. Send utterances
31	then read resulting session traces to ground your understanding of the
32	agent's behavior. Trace files reveal subagent selection, action I/O, and
33	LLM reasoning. DO NOT modify `.agent` files or action implementations without
34	this grounding. See [Validation & Debugging](references/agent-validation-and-debugging.md)
35	for trace file locations and diagnostic patterns.
36
37	4. Spec approval is a hard gate. Never proceed past Agent Spec
38	creation without explicit user approval.
39
40	5. Don't stall. After a step completes successfully, announce the
41	next step and start it. Do not wait for the user to say "what's next"
42	or "ok, continue." The only checkpoints that require explicit user
43	approval are: (a) Agent Spec approval, (b) the pre-Publish CHECKPOINT,
44	(c) any A/B branch the skill explicitly surfaces (e.g., Data Cloud
45	not provisioned during ADL setup). Long-running async work like ADL
46	indexing should run in the background while the skill continues with
47	work that doesn't depend on the result.
48
49	6. Draft-first lifecycle. During normal authoring, stay in draft iteration:
50	edit `.agent` + action implementations, validate, deploy, and preview as many
51	times as needed. Do NOT publish/activate by default. Publish + activate are
52	explicit release actions that require the user to confirm they are ready to
53	commit the current draft to metadata and expose it to end users.
54
55	7. Default agentic, pin with cause. Use the most agentic posture that meets
56	each subagent's requirement, and add deterministic controls only for
57	regulation/trust gates or observed failures. For detailed posture rules, see
58	[Posture & Determinism](references/posture-and-determinism.md).
59
60	8. No nested `if` or `else if`. Agent Script only supports flat `if`/`else` blocks. No `else if`, no `if` inside `else`, no `if` inside `if`. For multi-branch logic, use sequential `if` statements or compound conditions (`if A and B:`). Nested structures cause silent compile failures.
61
62	9. Action implementation is a user decision. During planning/spec work,
63	default new actions to `NEEDS STUB` placeholders. Always ask the user whether
64	they want to scan org/project for existing implementations and/or generate
65	new Apex/Flow/Prompt implementations before taking either path.
66
67	## Task Domains
68
69	Every task domain below has Required Steps. Follow verbatim, in order. The default path is: design -> draft implementation lo

People who liked this also viewed

6 picks

Type