agentleFS
seznamseznam/jailoc/.agents/skills/terminal-screenshots/SKILL.md

terminal-screenshots

Skill44 starsChanged 6 days agoMITView on GitHub

Regenerate the terminal screenshots of `jailoc traffic`, `jailoc strace`, `jailoc sessions` and `jailoc shell` output used in the docs and blog (docs/blog/img/*.png) without booting a VM. Use when adding, refreshing or restyling those images, or after the traffic/strace/sessions output format changes.

  • Installs packages

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. ricardodantas/skills★ 3same 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 repository5 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 1 is distinctive, and 6 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.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run from the repo root. W is a scratch dir outside the repo.

    line 26
  2. Very specific:

    Keep browsers under ~/.cache: /tmp can be a small tmpfs that Chromium does not fit in.

    line 49
  3. Very specific:

    If playwright install fails while extracting, download the zip URL it prints and unzip it into …

    line 50
  4. Distinctive:

    Check: every PNG by viewing it: colours present, no wrapped lines, font is JetBrains Mono (a fallback font means the font request failed).

    line 65
  5. Very specific:

    Keep event fields shaped like the real broker emits them: DNS type is TypeA (not A), names carry a trailing dot, flow reason is one of the …

    line 94
  6. Very specific:

    Read internal/netbroker and internal/guest when adding a new event kind.

    line 100
  7. Very specific:

    If the formatter output changes (a new column, different colours), just rerun the steps.

    line 102

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Terminal screenshots

The images are sample data rendered through jailoc's real formatters, so the layout and colours match the shipped CLI exactly; only the events are made up. A VM is never booted, so this works on hosts without KVM.

Pipeline: shots_test.go (Go, package cmd) feeds events through trafficFormatter / execFormatter from internal/cmd/inspectfmt.go and replicates the listSessionsVM table from internal/cmd/vmsession_unix.go; the jailoc shell transcript is written by hand. It writes transcripts traffic.ansi, strace.ansi, sessions.ansi and shell.ansi: lines starting with $ are prompts, the rest is command output. render.mjs turns each transcript into a 2x macOS-style dark terminal window PNG with Playwright/Chromium.

Steps

Run from the repo root. W is a scratch dir outside the repo.

  1. Generate ANSI (the driver is temporary; it lives in internal/cmd only while it runs):

    W=/tmp/opencode/shots
    cp .agents/skills/terminal-screenshots/shots_test.go internal/cmd/zz_shots_test.go
    SHOTS_DIR=$W TZ=Europe/Prague go test ./internal/cmd -run TestZZShots -count=1
    rm internal/cmd/zz_shots_test.go

    TZ=Europe/Prague is load-bearing: sample times are UTC and render as 14:31 CET. Done when $W holds four non-empty .ansi files and git status internal/ is clean.

  2. Prepare the renderer (once per machine):

    mkdir -p $W/render && cd $W/render
    npm i --no-audit --no-fund playwright@1.48.0
    cp <repo>/.agents/skills/terminal-screenshots/render.mjs .
    PLAYWRIGHT_BROWSERS_PATH=$HOME/.cache/pw-shots npx playwright install chromium

    Keep browsers under ~/.cache: /tmp can be a small tmpfs that Chromium does not fit in. If playwright install fails while extracting, download the zip URL it prints and unzip it into $PLAYWRIGHT_BROWSERS_PATH/chromium-<revision>/ (1.48.0 → chromium-1140).

  3. Render (needs network for the JetBrains Mono web font):

    cd $W/render
    PLAYWRIGHT_BROWSERS_PATH=$HOME/.cache/pw-shots node render.mjs $W <repo>/docs/blog/img

    Writes <name>.png per <name>.ansi. The first prompt titles the window; a transcript without a trailing newline ends with the cursor on its last line (the guest shell prompt); one whose last prompt ends in -f ends on a bare cursor (still following); others on a fresh $ prompt.

  4. Check every PNG by viewing it: colours present, no wrapped lines, font is JetBrains Mono (a fallback font means the font request failed).
  5. Clean up: rm -r $W. Leave ~/.cache/pw-shots for the next run or delete it.

Changing the story

The samples tell one consistent story that the blog captions in docs/blog/jailoc-2.md describe. Change events, PIDs or times together across all outputs and update the captions and alt texts to match.

  • traffic: OpenCode in workspace shop reaches api.anthropic.com, npm, GitHub over SSH, an internal GitLab allowed by exception 10.20.4.17/32, Debian over plain HTTP; then postgres.prod.corp.example is blocked by filtered_dns, 10.20.9.3 hits denied-range, the metadata IP hits hard-deny; finally an inbound connection to the published dev server port 5173.
  • strace: the processes behind that traffic, one tree under opencode (PID 812, parent 788 is jailoc's own hidden session process). Every recorded process gets an exit record; status 0 exits are hidden by the formatter, so only the failed psql (exit 2) and curl (exit 7, the broker resets denied connections) show.
  • sessions: one jailoc sessions listing five sessions with different harnesses (OmO, OpenCode, Copilot CLI, Claude Code, OpenCode --continue), attached and detached.
  • shell: jailoc shell in shop with Ubuntu's coloured bash prompt (agent@<workspace>:<host path>$; the VM hostname is the workspace name): id, the /run/secrets file secret, the npm registry reachable, the filtered_dns name not resolving (getent exit 2, the broker answers NXDOMAIN) and the metadata IP refused (curl: (7)).

Keep event fields shaped like the real broker emits them: DNS type is TypeA (not A), names carry a trailing dot, flow reason is one of the broker's reasons (public, exception <CIDR>, denied-range, hard-deny, …) and session IDs are 6 hex characters. Agent command lines are deliberately short (binary name, plus --continue): the real ones also carry unattended-mode and plugin flags (--dangerously-skip-permissions, --settings …), which only add noise. Read internal/netbroker and internal/guest when adding a new event kind.

If the formatter output changes (a new column, different colours), just rerun the steps; if the sessions table changes, update the hand-made copy in shots_test.go.

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.