agentleFS
Fighter90Fighter90/career-ops-ui/.claude/skills/hermes-bridge/SKILL.md

hermes-bridge

Skill99 starsChanged 8 days agoMITView on GitHub

Deploy career-ops-ui to a cloud server and/or bridge it to Telegram through a Nous Research Hermes agent. Walks the documented steps, checks prerequisites (keys, endpoint reachability via the SSRF-safe path, Node version), and never writes secrets to disk or logs. Trigger when the user says "connect career-ops to Telegram", "Hermes bridge", "deploy career-ops to a cloud box", "wire the pipeline to Telegram via Hermes", or asks how to run the UI on a server.

  • Reads credentials

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. zedarvates/botte-secrete★ 2same name
    2. zedarvates/frugal-skills★ 1same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository7 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Verify, and report any gap instead of proceeding:

    line 51

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

hermes-bridge — cloud deploy + Telegram-via-Hermes

Single source of truth: docs/integrations/HERMES.md. This skill operationalizes that guide — it does not restate it. If the two ever disagree, the doc wins; fix the skill to match. Read the doc first, then walk the steps below.

Status (v1.151.0): the LLM-provider path — Shape A — is WIRED. Hermes's hermes gateway exposes an OpenAI-compatible POST /v1/chat/completions, so career-ops-ui already integrates it as a provider: set HERMES_API_KEY (+ optionally HERMES_BASE_URL, HERMES_MODEL) in #/config. What this skill still does is the operator how-to: (a) deploy the existing UI to a cloud box and (b) prepare the Telegram-via-Hermes bridge — neither is a shipped app feature. It must still never route a secret to disk/log/Telegram or weaken a security header.

Non-negotiable guardrails (apply to every step)

  1. Secrets never touch disk or logs. Provider keys and the Telegram bot token go in the parent .env (gitignored) or Hermes's own config. Do not echo them, do not write them into a note, do not paste them into a commit.
  2. The parent career-ops project is read-only except the app's existing user-action write-through. This skill deploys the viewer; it does not edit ../cv.md, ../config/, ../reports/, etc.
  3. Reachability checks use the SSRF-safe path only. To test whether a Hermes / Nous Portal endpoint is reachable, go through isValidJobUrl() + safeGet (the same guard the app uses) — never a raw fetch/curl to an unvalidated user-supplied URL, and never to loopback or file://.
  4. The security envelope survives the move off 127.0.0.1 — CSP (no unsafe-inline/unsafe-eval, frame-ancestors 'none'), the SSRF guard, the stripDangerousMarkdown() / UI.md() markdown boundary, and no-secrets-in-logs all stay intact. Do not relax a header to make remote deployment easier.

Step 0 — The provider is already wired (Shape A, v1.151.0)

The scoping spike is done: Hermes's API Server (hermes gateway) is OpenAI-compatible (POST /v1/chat/completions at http://127.0.0.1:8642/v1, Bearer API_SERVER_KEY, streaming, GET /v1/models), so Shape A shipped — career-ops-ui reaches it via the shared runOpenAICompatible client. To use it, the user just sets HERMES_API_KEY in #/config and starts hermes gateway; no new code is needed. This skill's job from here is the deployment + Telegram how-to below. (Shape B — a bespoke agent-runtime relay — was not needed.)

Step 1 — Prerequisite check (deployment)

Verify, and report any gap instead of proceeding:

  • Node ≥ 18 on the target box (node -v).
  • Parent career-ops present with this repo at career-ops/web-ui/.
  • Provider key(s) in the parent .env (presence only — never print the value).
  • The app boots and /api/health is green on loopback before exposing it.

Step 2 — Cloud deploy (per HERMES.md §2)

Walk the user through: bind to loopback + reverse proxy (nginx/Caddy) terminating HTTPS → 127.0.0.1:4317; a systemd/pm2 unit under a non-root user with Restart=on-failure; and — only if HOST=0.0.0.0 is required on a container network — confirm the rate-limit / SSRF / path-sanitize hardening (a no-op on loopback) is active. Never hand the user a config that exposes 0.0.0.0 to the public internet directly.

Step 3 — Telegram-via-Hermes (per HERMES.md §3)

Stand up the Hermes agent, connect its Telegram channel with a @BotFather token that lives in Hermes's config (career-ops-ui never sees it). Pick the bridge direction — push (career-ops-ui → Hermes → Telegram) or tool-call (Hermes calls a read-only UI endpoint) — and apply the §3.2 "what NOT to expose" list: no CV text, no salary numbers, no raw report bodies, no keys, no internal URLs. Send the minimum useful summary + a link the authenticated user opens themselves.

Step 4 — Honesty check before you finish

Confirm you did not: weaken a security header, expose 0.0.0.0 directly to the internet, or route a secret to disk/log/Telegram. If the user only wanted the LLM-provider integration, that already shipped (Shape A) — point them at #/config (HERMES_API_KEY) and docs/integrations/HERMES.md.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.