| name | herdr |
| description | Use this skill when running inside Herdr to coordinate workspaces, tabs, panes, sibling agents, server/test/log panes, agent-to-agent messages, or multi-agent delegation. Send a lightweight herdr-msg (reply-to + task), then stop and end the turn. Do not poll siblings and do not block-wait for replies; handle the peer message when it is injected into your pane. |
Herdr
Herdr is a terminal workspace manager for coding agents. Use it to inspect sibling panes, delegate to other agents, run servers or tests beside your current pane, and coordinate work without blocking.
For exact CLI syntax and less common workspace/tab/worktree operations, read references/command-reference.md.
Preconditions
test "${HERDR_ENV:-}" = "1"
command -v herdr
If HERDR_ENV is not 1, do not control live panes unless the user asked you to edit this skill or discuss Herdr. You may still read files and explain commands.
Operating Rules
- Discover first:
herdr agent list, herdr pane list.
- Treat IDs as live-session handles. Re-read them after layout changes.
- Prefer
HERDR_PANE_ID as your pane handle when set.
- For agent task handoff use
scripts/herdr-msg or herdr pane run (not herdr agent send alone — it does not press Enter).
- Read a target once before sending, only to avoid interrupting a prompt/approval.
- After send: stop. Do not poll the target. Do not
herdr wait / --wait-reply for normal chat. End your turn. When a reply is injected into your pane, handle that inbound message.
- Do not use
herdr wait agent-status ... --status done for multi-agent coordination.
Default Loop
- Discover agents/panes.
- Read the target once if needed.
- Send a short request with
scripts/herdr-msg (header carries reply-to + task).
- On
state=delivered: end the turn (or do unrelated local work that does not depend on the reply).
- When an inbound reply appears in your pane, integrate it.
After herdr-msg (mandatory)
- Non-zero exit → send/resolve failed; fix and resend. Do not poll.
state=delivered (exit 0) → stop. Do not read the target in a loop. Do not block-wait.- End your turn. The wake-up is the peer injecting a message into
reply-to (your pane).
- Polling the target after send is a protocol violation.
Lightweight agent messages
Keep it human and thin. One header line + body:
[herdr-msg reply-to:<your-pane-id> task:<short-label>]
<short actionable message>
Optional fields the helper may add: from:<name>, kind:reply / kind:update (requests omit kind).
| Field | Role |
|---|
reply-to | Required. Where the peer should answer. |
task | Human thread label (not a crypto id). |
from | Optional sender label. |
kind | Optional; default is request. |
Not required for normal chat: msg_id, sentinels, ack-of, status codes in the header, waiting recipes.
Send (default)
scripts/herdr-msg codex --task auth-review <<'MSG'
Review src/auth.ts for auth bypasses and missing tests.
Reply to reply-to with DONE or BLOCKED findings.
MSG
Reply
scripts/herdr-msg "$REPLY_TO" --task auth-review --kind reply <<'MSG'
DONE: Checked src/auth.ts. No bypass found. Missing refresh-token expiry test.
MSG
Manual fallback
SELF=$HERDR_PANE_ID
TARGET=$(herdr agent get codex | sed -nE 's/.*"pane_id":"([^"]+)".*/\1/p')
herdr pane run "$TARGET" "[herdr-msg reply-to:$SELF task:auth-review]
Review src/auth.ts. Reply to reply-to with DONE or BLOCKED."
Receipt
Tiny key=value lines: ok, state, target, target_pane, reply_to, task, kind, hint.
Treat state=delivered as success. The hint says to end the turn.
Waiting Policy
Agent chat: do not wait. Send, end turn, handle inbound reply later.
Use herdr wait output only for non-agent processes (dev servers, tests) where you can name a concrete marker such as ready or FAIL:
herdr wait output "$PANE" --match "ready" --timeout 30000
herdr pane read "$PANE" --source recent --lines 40
Only use herdr agent wait / wait agent-status when the user explicitly needs UI status or you are attaching/taking over an agent.
Delegation Pattern
herdr agent list- Read target once.
scripts/herdr-msg <target> --task <label> <<'MSG' ... MSG- On delivered: end turn (or unrelated local work). No poll, no block-wait.
- When replies arrive on your pane, integrate; follow up only for gaps.
Server And Test Panes
SELF=${HERDR_PANE_ID:-$(herdr pane list | sed -nE 's/.*\{[^{}]*"focused":true[^{}]*"pane_id":"([^"]+)".*/\1/p')}
PANE=$(herdr pane split "$SELF" --direction right --cwd "$PWD" --no-focus | sed -nE 's/.*"pane_id":"([^"]+)".*/\1/p')
herdr pane run "$PANE" "npm run dev"
herdr wait output "$PANE" --match "ready" --timeout 30000 || true
herdr pane read "$PANE" --source recent --lines 50
For tests, wait for framework markers (passed, failed, FAIL, …) or run tests in your own session.
Gotchas
pane run submits with Enter; agent send does not.recent-unwrapped is better than recent when matching command output.- Herdr sets
HERDR_ENV=1, HERDR_PANE_ID, HERDR_SOCKET_PATH in panes.
- Long diffs/logs: write a file path into the message; do not paste huge blobs into the peer prompt.
- Request-to-self is refused by default (
--allow-self to override).
- Keep headers one line; put status and findings in the body (
DONE: … / BLOCKED: …).
