salesforce-agent-skills
AGENTS.md2 starsChanged 31 days agoMITView on GitHub
This repository ships two agent skills for Salesforce automation work. Each folder under skills/ is the single source of its skill's content. Codex discovers them through the .agents/skills/ symlinks and the repo-local marketplace in .agents/plugins/, and Claude Code loads them through the plugin manifest in .claude-plugin/. salesforce-flow-modernizer builds, audits, and migrates Salesforce Flows with current platform elements instead of loop-and-decision patterns. It carries: - SKILL.md, with the workflow, a decision table from requirement to element (with metadata tag, minimum org…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 33,315 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository3 other files it gives its agents.
The file
We split this AGENTS.md into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 28
Before any design, the skill reads references/org-conventions.md, and a filled-in decision there outranks the skill's defaults.
- Distinctive: line 58
Never assert Salesforce feature availability, limits, GA status, or deprecations from memory when the answer changes the design or an audit …
- Distinctive: line 83
Follow CONTRIBUTING.md. It holds the rules for content, code comments and examples, the scratch-org deploy every example change needs, the …
≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AGENTS.md
This repository ships two agent skills for Salesforce automation work. Each
folder under skills/ is the single source of its skill's content. Codex
discovers them through the .agents/skills/ symlinks and the repo-local
marketplace in .agents/plugins/, and Claude Code loads them through the
plugin manifest in .claude-plugin/.
Skills
salesforce-flow-modernizer builds, audits, and migrates Salesforce Flows
with current platform elements instead of loop-and-decision patterns. It
carries:
SKILL.md, with the workflow, a decision table from requirement to element (with metadata tag, minimum org API version and example file), and the escape conditions under which a legacy pattern is correct.references/, with one file per domain (collections, screens, record-triggered, integration and security, AI and agents, legacy migration, delivery) plusorg-conventions.md. The agent reads only the domains a task touches.examples/, with one deploy-verified.flow-meta.xmlper pattern. Each has a header comment naming the pattern, what it replaces, and its shape rules.scripts/check-flow.mjs, a dependency-free checker for legacy patterns and for shapes the platform rejects at deploy time.
Before any design, the skill reads references/org-conventions.md, and a
filled-in decision there outranks the skill's defaults.
salesforce-automation-mapper lists all the automation on one object in a
local sfdx project and the order it runs in, without changing anything. It
carries:
SKILL.md, with the workflow and the output format.references/save-order.md, with the stages of a save and the four order rules, each with its source.scripts/map-automation.mjs, a dependency-free script that lists the object's flows, processes, workflow rules and Apex triggers and the fields each writes. It orders nothing.
It reads the conventions file from
../salesforce-flow-modernizer/references/org-conventions.md, so the two
skills sit next to each other.
A platform fact both skills state lives in salesforce-flow-modernizer
between <!-- shared:<id> --> and <!-- /shared:<id> --> markers. The
mapper holds a copy, and the smoke test fails while a copy differs.
Invocation
In Codex CLI or the IDE extension, list skills with /skills, mention one
with $salesforce-flow-modernizer or $salesforce-automation-mapper, or
rely on implicit invocation through the skill descriptions.
Verification rule
Never assert Salesforce feature availability, limits, GA status, or deprecations from memory when the answer changes the design or an audit finding. Check the target org's API version and the official documentation on help.salesforce.com and developer.salesforce.com.
Platform notes
- Codex offers the plugin's hook for trust. On
apply_patchedits the hook reads the patch text intool_input.commandand gives the Flow reminder when the patch adds, updates or moves a file into amain/default/flows/folder. Claude Code sendstool_input.file_path. - The hook script needs Node.js on the PATH. It's advisory, which means it adds a reminder to the session and never blocks a tool call.
Keeping the skills current
To check the skills against a new Salesforce release, use the maintainer skill
release-review in .claude/skills/ (Codex finds it through
.agents/skills/). It doesn't ship in the plugin. maintenance/release.json
records the release the skills were last reviewed against, and the
"Release reviews" section of CONTRIBUTING.md explains the scripts it uses.
Editing this repository
Follow CONTRIBUTING.md. It holds the rules for content,
code comments and examples, the scratch-org deploy every example change
needs, the smoke test to run before a commit, and the commit and pull request
conventions. The behavioural eval suite lives in evals/, and
evals/README.md explains it.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

