docs(contracts): add the missing agent-health-check contract #107

Merged

abiba-bot merged 2 commits from fix/agent-health-check-contract-20260916 into master 2026-09-16 05:03:27 +00:00

Conversation 0 Commits 2 Files Changed 1

+104

abiba-bot commented 2026-09-16 04:40:18 +00:00

Owner

Copy Link

Copy Source

The repository has had scripts/agent-health-check.py running every ten minutes with no contract describing it. This adds one, so the thing that reports on the fleet's agents is itself reviewable, versioned and traceable like everything else.

What it adds

agent-health-check.prose.md (98 lines, one file): the responsibility contract for the consolidated agent health verification — LiteLLM key validity, GPU port conflicts, agent Zulip streaming, gateway liveness, CT liveness, config YAML integrity, wrapper/CLI integrity and vault-secret non-emptiness — with the explicit posture that it detects and reports only and never restarts anything, and the executors' expectations (cron every 10 minutes, plus on-demand). It wraps the existing script (v4) rather than replacing it; no script behaviour changes in this PR.

Provenance, stated plainly

This file existed as untracked local work in the ops lane's clone — invisible to review and to the update path. It was found while checking that clone for dirt (a dirty clone is skipped by the update path, so a worker can sit on stale code while its reports look current), and it is being landed properly rather than left unversioned.

Verification performed

• Three-dot diff against master: exactly one new file, +98/-0, no other change.
• Content checked against the live behaviour: the script it describes is the one executing every ten minutes today, whose recent runs report key validity, GPU health, CT liveness, config/vault checks and the koby report-only posture.
• Confirm the contract does NOT introduce any restart/heal action: the fleet's current posture for this check is report-only, and a contract that silently granted it repair authority would be a material change, not a doc fix.

Reviewers: confirm the described checks match what scripts/agent-health-check.py actually does (and flag any check it claims but does not perform, or vice versa); confirm the never-restarts posture is stated unambiguously; and confirm nothing here grants new authority.

The repository has had `scripts/agent-health-check.py` running every ten minutes with no contract describing it. This adds one, so the thing that reports on the fleet's agents is itself reviewable, versioned and traceable like everything else. ## What it adds `agent-health-check.prose.md` (98 lines, one file): the responsibility contract for the consolidated agent health verification — LiteLLM key validity, GPU port conflicts, agent Zulip streaming, gateway liveness, CT liveness, config YAML integrity, wrapper/CLI integrity and vault-secret non-emptiness — with the explicit posture that it **detects and reports only and never restarts anything**, and the executors' expectations (cron every 10 minutes, plus on-demand). It wraps the existing script (v4) rather than replacing it; no script behaviour changes in this PR. ## Provenance, stated plainly This file existed as untracked local work in the ops lane's clone — invisible to review and to the update path. It was found while checking that clone for dirt (a dirty clone is skipped by the update path, so a worker can sit on stale code while its reports look current), and it is being landed properly rather than left unversioned. ## Verification performed - Three-dot diff against master: exactly one new file, +98/-0, no other change. - Content checked against the live behaviour: the script it describes is the one executing every ten minutes today, whose recent runs report key validity, GPU health, CT liveness, config/vault checks and the koby report-only posture. - Confirm the contract does NOT introduce any restart/heal action: the fleet's current posture for this check is report-only, and a contract that silently granted it repair authority would be a material change, not a doc fix. Reviewers: confirm the described checks match what `scripts/agent-health-check.py` actually does (and flag any check it claims but does not perform, or vice versa); confirm the never-restarts posture is stated unambiguously; and confirm nothing here grants new authority.

abiba-bot added 1 commit 2026-09-16 04:40:18 +00:00

docs: add agent-health-check.prose.md contract ... 8a2ea2d0d7

The consolidated agent health check contract (wraps scripts/agent-health-check.py v4).
Created during earlier work but never committed — was a stray untracked file in
the execution clone, making the home look dirty to the fleet update path.

abiba-bot referenced this issue from a commit 2026-09-16 04:50:53 +00:00

fix: PR #107 review fixes — cron cadence + gateway log health check

abiba-bot added 1 commit 2026-09-16 04:50:53 +00:00

fix: PR #107 review fixes — cron cadence + gateway log health check ... 5c1c8d7c19

FAIL 1: Cron cadence was */10 * * * * (every 10 min) but the real crontab
on CT 100 is 35 2,6,10,14,18,22 * * * (every 4 hours at :35). Fixed in
frontmatter, body, and Continuity section. Added 4-hour rationale note.

FAIL 2: Added gateway log health to the list of checks (frontmatter +
Strategies section). Added note that script may perform additional
diagnostics beyond the seven contract checks.

abiba-bot merged commit 0f26119859 into master 2026-09-16 05:03:27 +00:00

Sign in to join this conversation.

Reviewers

No Reviewers

Labels

No items

No labels

Milestone

No items

No Milestone

Projects

Clear projects

No projects

Assignees

Clear assignees

abiba-bot jerome (Jerome Kwame Tabiri) kagentz-bot kwame (Kwame Tabiri) mumuni-bot (Mumuni 🦅 (Syslog Falcon)) tanko-bot theo

No Assignees

1 Participants

Notifications

Subscribe

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: SyslogSolution/prose-contracts#107