agentleFS
colinc86colinc86/gloss/.claude/skills/verify-transport/SKILL.md

verify-transport

Skill0 starsChanged 44 days agoMITView on GitHub

Live end-to-end check of one delivery transport (herdr, mcp, opencode, tmux, clipboard) against the real tools installed on this machine or a remote host, reporting what was sent, what arrived, and what to fix. Invoke as /verify-transport <name> [host].

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    opencode: start opencode serve --port 4096 in the background, GET /session, create a session if needed, POST /session/:id/message with a …

    line 18
  2. Very specific:

    tmux: create a detached session gloss-verify, load-buffer + paste-buffer -p + Enter, capture-pane -p to confirm the packet text, then …

    line 19

≈ “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

Verify the $0 transport end to end on host $1 (default: local). Use the herdr-api, agent-integrations, and context-packet skills for the exact protocol. Follow the recipe for the named transport, then report: the exact packet sent, evidence it arrived (pane read-back, hook stdout, HTTP response, clipboard contents), and any fix needed in the app code. Never leave test panes, buffers, or temp files behind.

Recipes:

  • herdr: herdr status must show the server running. herdr agent list --json; pick an idle agent pane or, if none, split a pane and start claude in it. Send a two-note packet with agent.prompt over the socket (nc -U), wait up to 60s, read back with pane.read source=recent, confirm both file headers appear. Close any pane you created.
  • mcp: launch the app (run-app skill) with --reset-db, create a note via the URL scheme and send it with the Clipboard target (back the clipboard up first; agents only see sent notes, so list_pending_notes must not list it while it is a draft), then curl the MCP endpoint: initialize, tools/list, tools/call list_pending_notes, tools/call resolve_note. Confirm the note's state flips in the SQLite db under $TMPDIR.
  • claude-hook / codex-hook: removed (issue #5). Check instead that Settings › Agents Install takes an old gloss-notes.sh entry out of ~/.claude/settings.json / ~/.codex/hooks.json in a temporary HOME, with a backup. Former recipe, for reference: run the hook script directly with a synthetic UserPromptSubmit JSON on stdin and a pending note in a throwaway db; confirm it prints hookSpecificOutput.additionalContext under 10,000 chars. Then confirm ~/.claude/settings.json contains the hook entry without having clobbered other hooks.
  • codex-hook: same with ~/.codex/hooks.json and [features] hooks = true in ~/.codex/config.toml.
  • opencode: start opencode serve --port 4096 in the background, GET /session, create a session if needed, POST /session/:id/message with a text part and an image part, confirm 200 and a reply. Kill the server.
  • tmux: create a detached session gloss-verify, load-buffer + paste-buffer -p + Enter, capture-pane -p to confirm the packet text, then kill-session.
  • clipboard: send to clipboard, pbpaste and diff against the expected packet.

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.