feature: support unattended agent workloads — headless agent runs with no human present
@esnible is already working on this.
Since Sep 10, 2026.
Assessment
This issue has not been assessed yet.
Description
Feature Description
Cortex assumes a human today: abctl claude-code enable asks before editing settings, and you
only see anything if abctl observe is open in a terminal. Agents increasingly run with nobody
watching — claude -p inside a script, CI jobs, cron, server-side and sandboxed agents. Those
are precisely the runs where a mistake goes unnoticed, so they are the ones that most need recording.
Proposed Solution
- Non-interactive enable / disable / status: no prompts, explicit
--yes, exit codes a
script can branch on (exitDeclined = 3already exists — keep the contract documented) - Configuration and credentials from the environment; no interactive login, no keychain prompt
- Events from an unattended run are captured and attributable after the fact, with no
abctl observerunning at the time - A headless run gets its own session identity so its tokens and cost are separable from
interactive work - That run's totals readable once it has finished
- Works with no TTY and no
TERM - Failures land in a log the caller can find, not only on a screen nobody is watching
- Documented for CI and sandbox use
Additional Context
Part of the Cortex v0.9.0 release. Depends on session differentiation. The release smoke tests
assert events through this capture path.
- Dominant language
- Go
- Stars
- 13
- Forks
- 40
- Avg merge
- 12h 17m
- Merged PRs (30d)
- 156
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from rossoctl/cortex
-
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
-
nice to have
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
-
Difficulty 2/5 1-3 hours Newbie friendliness 88/100
-
stale
Difficulty 2/5 1-3 hours Newbie friendliness 88/100
-
enhancement
Difficulty 5/5 Over a week Newbie friendliness 35/100
Similar issues
-
optimization optimization:agents-md-curator
Difficulty 2/5 1-3 hours Newbie friendliness 86/100
githubnext/gh-aw-cao#13143 ·
-
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
blinklabs-io/bursa#904 ·
-
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
yanet-platform/ipfw-go#129 ·
-
bug confmap/provider/googlesecretmanagerprovider needs triage
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
open-telemetry/opentelemetry-collector-contrib#51273 · 2 comments ·
-
bug: AI Gateway client filter lists "Unknown" twice when NULL and literal Unknown clients coexist Openbug
Difficulty 2/5 1-3 hours Newbie friendliness 90/100