Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add freshworks-developers/fw-dev-tools --skill fw-publishgit clone --depth 1 https://github.com/freshworks-developers/fw-dev-toolsWrote 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/skills/freshworks-developers/fw-dev-tools/fw-publish)<a href="https://agentmods.dev/skills/freshworks-developers/fw-dev-tools/fw-publish"><img src="https://agentmods.dev/badge/skills/freshworks-developers/fw-dev-tools/fw-publish/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/skills/freshworks-developers/fw-dev-tools/fw-publish"><img src="https://agentmods.dev/badge/skills/freshworks-developers/fw-dev-tools/fw-publish.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 | $0.00170 | $0.08248 |
| Opus 5.5 | $0.00068 | $0.03299 |
| Sonnet 5.5 | $0.00034 | $0.01650 |
| Haiku 4.5 | $0.00017 | $0.00825 |
Grade C, and why
fw-publish scanned grade C with 2 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 1mo 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
rm -rf /tmp/fw-repack && mkdir -p /tmp/fw-repack && unzip -q -o 'dist/<app>.zip' -d /tmp/fw-repack Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
**Sandbox / agent egress:** Publish **does not work** in **sandbox** (or equivalent) modes that block MCP traffic to `https://mcp.freshworks.dev/mcp` or outbound HTTPS (including **`curl`** PUT to the app-upload URL). So How it starts
The opening of the file, as written. The whole thing — 418 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Entry contract
- Invoked by: controller after fw-review pass and explicit user consent to publish
- Preconditions: fw-review completed; MCP auth confirmed (
list_custom_appssmoke test); non-sandbox network for upload - Returns:
done(submit/version created) |blocked(auth/network/review missing) |escalate(repeated upload failures) - On complete: hand control back to the controller — update session publish block via
session-write.sh; do not start new workflows - Session: read
publish.tracking_idfrom session; writepublish.tracking_id,step=publishedafter successful submit
Publish (Platform 3.0 custom app)
MANDATORY PREREQUISITE: Run fw-review skill before publishing to ensure marketplace compliance. The review checks iparams, frontend files, security patterns, and generates a structured audit report. Do not proceed with publishing until review passes.
Goal: Take any Platform 3.0 app folder (with manifest.json), produce dist/*.zip, and publish it to the Freshworks Marketplace via MCP tools. Currently, only test state is supported.
Sandbox / agent egress: Publish does not work in sandbox (or equivalent) modes that block MCP traffic to https://mcp.freshworks.dev/mcp or outbound HTTPS (including curl PUT to the app-upload URL). Some cloud or CI agent runtimes send HTTPS through a proxy, gateway, or assumed IAM role that is not allowed to PUT to the Marketplace app-upload bucket (fa-*-app-uploads…): S3 returns 403 / AccessDenied even when the presigned URL is valid. Do not run this playbook in a sandboxed agent or restricted shell. If the user’s client offers sandbox vs full network access, choose non-sandbox / allow network for publish. On 403 PUT after following step 8 exactly, ask the user to run the same script command on their local machine (see step 8).
Agent playbook (MCP tools)
Tool names and parameters match Freshworks openai-server (references/openai-server-mcp-tools.md). Hosted MCP: https://mcp.freshworks.dev/mcp.
What ships with it
16 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- .claude-plugin/plugin.json 636 B
- .cursor-plugin/plugin.json 1.4 KB
- .gitignore 138 B
- README.md 3.4 KB
- references/deprecated-mcp-build-tools.md 2.7 KB
- references/openai-server-mcp-tools.md 5.6 KB
- references/templates/claude-mcp-setup.json 175 B
- references/templates/cursor-mcp-setup.json 175 B
- references/templates/custom-app-limit-warning.txt 244 B
- references/templates/developer-feedback-comment-prompt.txt 87 B
- references/templates/developer-feedback-rating-prompt.txt 139 B
- references/templates/downgrade-warning.txt 257 B
- references/templates/engines-mismatch-prompt.txt 526 B
- references/templates/stuck-version-warning.txt 302 B
- scripts/repack-app-zip.sh 813 B runs code
- scripts/upload-app.sh 1.8 KB runs code
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.
- 1mo ago Changed · +8 lines a951d506f8ce
- 1mo ago Changed 38dccc8ff91e
- 1mo ago First seen · 410 lines · 170 tokens per session scan C bcf83ee588bb
fw-publish is a skill published in the GitHub repository freshworks-developers/fw-dev-tools (5 stars, last pushed 1mo ago), licensed MIT. It adds 170 tokens to every session and 8,248 once invoked, about $0.0007 per session on Opus 5.5. A static security scan graded it C with 2 findings (recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
dependency-vulnerability
OWASP A09 - Using Components with Known Vulnerabilities. Use this skill when auditing dependencies, updating packages, or reviewing security advisories. Activate when: npm audit, dependency check, vulnerable package, CVE, security advisory, outdated packages, supply chain, package vulnerability, Dependabot, Snyk.
ai-amazon-brand-analytics
An Amazon Brand Analytics assistant for working with Amazon brand-analysis tasks. The description does not provide enough detail about its exact data or outputs.
ai-amazon-international-listings
An Amazon localization assistant for checking whether a product listing is written correctly for an overseas market. Localization means adapting language and presentation to a specific country or region.
ai-amazon-inventory-management
An Amazon inventory-review assistant. Inventory means the products a seller has available to sell.
ai-amazon-listing-images
An Amazon listing-image review assistant. A listing is the product page shoppers see on Amazon, including its images.
ai-amazon-product-bundling
An Amazon product-bundling assistant. Product bundling means offering multiple products together as one package.