hc
Skill41 starsChanged 6 days agoApache-2.0View on GitHub
Run work through herdr-core (hc): servers, builds, test suites, CI watches and anything slow; commands and files on other machines; another agent or model (Claude Code, Codex, pi) as a peer; messages between agent sessions. Use whenever $HC_TERMINAL_ID is set and work would block your shell, outlive your turn, or leave this machine.
In this repository7 other files it gives its agents.
- AGENTS.mdAGENTS.md
- chainsSkill
- codebase-orientationSkill
- concept-diagramsSkill
- datadog-pupSkill
- diagnoseSkill
- validation-reviewSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 22
Never &, nohup, setsid, disown, or a foreground sleep to wait: hc keeps the process visible, outlives your turn, and hc wait returns the …
- Very specific: line 52
Never hc server stop, and never restart the core: every terminal on it dies.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
hc
You run in an hc terminal ($HC_TERMINAL_ID is your address). The person watches every hc
terminal in the Herdr app and can take any of them over. Full manual: hc --skill; read
it before anything not covered here.
What goes where
| Work | Use |
| A sub-task inside your turn: research, a review pass, a workflow | your own subagents / workflows, not hc |
| A server, build, test suite, CI watch, anything slow | hc run --bg, then hc wait |
| Another agent or model as a peer you or the person can steer | hc run --bg <agent> (see peers.md) |
| A command or file on another machine | hc exec / hc file / hc run --on (see remote.md) |
| Talking to another agent's session | hc send, hc wait, hc read |
Never &, nohup, setsid, disown, or a foreground sleep to wait: hc keeps the
process visible, outlives your turn, and hc wait returns the moment it's done.
Recipes
# slow command: you stay free while it runs
hc run --bg --rm --no-notify --name build cargo build --release
hc wait build # exit code; --rm removes it after
# server: start, wait until ready, use it, stop it
hc run --bg --name web --cwd app bun dev
hc wait web --output "ready" # matches output printed after the wait starts
hc read web --lines 50
hc kill web; hc rm web
# several at once
hc run --bg --name unit cargo test
hc run --bg --name lint cargo clippy --all-targets
hc wait --children any # next one to finish; repeat or use --children allhc waittimes out after 180 s by default: that's a check-in, not a failure; wait again.hc runwithout--bgstreams output and returns after 180 s with[hc] still running as tN: wait on that address, don't start it twice.- Clean up what you started (
hc kill,hc rm).hc lsshows the person's other work too: never touch terminals you didn't start.
Gotchas
- Never
hc server stop, and never restart the core: every terminal on it dies. - Starting
hcwhen no core is running starts one with your shell's environment, and every later terminal inherits it. On the person's own machine, checkhc lsworks before running anything else. - From inside an hc terminal,
hc runchildren end when you end; add--outlive-parentfor something that must keep running. hc sendwraps the text as a message, so/compactor any slash command arrives as chat. Type it as keys:hc keys <t> / c o m p a c t, thenhc keys <t> enter.- Codex under hc needs
--no-daemon.
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
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.

