forcedotcom/sf-skills is a collection of reusable instructions that help AI coding agents build and work with Salesforce applications and platform features. Developers use it for Agentforce agents, Lightning apps, Flow, Apex, SOQL, Lightning Web Components, and related Salesforce configuration. The catalogue entries are the repository’s skills, plugins, commands, MCP servers, and agents for AI tools that support skills.
Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add forcedotcom/sf-skills/plugin install salesforce-developmentWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/forcedotcom/sf-skills/feedback)<a href="https://agentmods.dev/commands/forcedotcom/sf-skills/feedback"><img src="https://agentmods.dev/badge/commands/forcedotcom/sf-skills/feedback/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/forcedotcom/sf-skills/feedback"><img src="https://agentmods.dev/badge/commands/forcedotcom/sf-skills/feedback.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00023 | $0.01065 |
| Opus 5.5 | $0.00009 | $0.00426 |
| Sonnet 5.5 | $0.00005 | $0.00213 |
| Haiku 4.5 | $0.00002 | $0.00106 |
Grade A, and why
feedback scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 28d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Collect a quick satisfaction rating about this plugin as usage telemetry. Like every other event this plugin captures, the rating is buffered locally first, then sent later — at session end, through your connected org, the same pipeline as the rest of this plugin's telemetry — subject to your telemetry settings. This command never sends free text, paths, code, or org data anywhere; the rating is the only thing that goes out.
No matter which outcome below this ends on — ineligible, recorded, skipped, error, or given up after retries — finish with one more line: "Want to share more than a number? Open an issue: https://github.com/forcedotcom/sf-skills/issues/new"
-
Check eligibility first, before asking anything:
"${CLAUDE_PLUGIN_ROOT}/scripts/sf-context" feedback-eligibilityThis prints
{"eligible": true}or{"eligible": false, "reason": "<reason>"}and captures nothing by itself. Ifeligibleisfalse, do NOT ask for a rating — tell the user plainly why none would be recorded right now, then stop:telemetry_disabled→ "Your rating wouldn't be recorded because telemetry is turned off."not_sf_project→ "Your rating wouldn't be recorded because this isn't a Salesforce DX project — feedback capture is scoped to project work like the rest of this plugin's telemetry."- any other reason (e.g.
unavailable) → "Your rating can't be recorded right now."
-
If eligible, ask for a 1-5 rating (1 = not helpful at all, 5 = extremely helpful), in the chat as a plain question — do NOT use a structured multiple-choice/selection tool for this: a 1-5 scale is 5 explicit values, at or past what that kind of tool allows as explicit options, so reaching for it here can fail outright. If the user's arguments ($ARGUMENTS) already state a number in that range, use it directly and skip the question — no need to narrate that it was already supplied, just move straight to recording it.
Validate before moving on: if $ARGUMENTS or the user's chat reply isn't an integer 1-5 (e.g.
0,-100, "not sure"), don't send it tofeedback-record— ask again for a valid number instead. Give up after 3 total attempts (the initial ask plus 2 re-asks) — don't loop forever. If the 3rd attempt is still invalid, say "No worries — skipping the rating for now. Run/salesforce-development:feedbackagain anytime." and stop without callingfeedback-record. Same if the user says something like "never mind"/"skip"/"cancel" at any point — stop immediately, don't spend the remaining attempts. -
Record the rating with a single literal command — no pipe, no shell variable expansion, substituting only the actual numeric rating for
<rating>:"${CLAUDE_PLUGIN_ROOT}/scripts/sf-context" feedback-record <rating>This reads the session id itself and is consent-gated and fail-silent like every other telemetry capture in this plugin (it honors
SF_DISABLE_TELEMETRY/DO_NOT_TRACKand the/salesforce-development:telemetry offsetting). It prints exactly one of:{"result": "recorded"}— the rating was actually written to the local telemetry buffer.{"result": "skipped", "reason": "<reason>"}— nothing was written (same reason vocabulary as step 1, plus"unavailable"for a transient failure).{"result": "error", "reason": "invalid_rating"}— the argument wasn't a bare integer 1-5.
-
Tell the user exactly what that result was — never assume success:
recorded→ one concise line, e.g. "Thanks — your rating of 4/5 was recorded."skipped→ a short, accurate explanation using the reason (same wording as step 1's mapping, or "Your rating wasn't recorded — telemetry is unavailable right now." forunavailable).error→ should be rare, since step 2 already validates the number before it ever reaches this command — but if it happens anyway, say "That wasn't a valid 1-5 rating, so nothing was recorded." and stop; don't silently substitute a number and resubmit on your own.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 28d ago First seen · 66 lines · 23 tokens per session scan A 2172129b3a1b
feedback is a command published in the GitHub repository forcedotcom/sf-skills (1,104 stars, last pushed yesterday), licensed Apache-2.0. It adds 23 tokens to every session and 1,065 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-23.
Other commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.
speckit.bug.fix
Apply the remediation from a bug assessment and record what was changed.