oncall-engineer

oncall-engineer is an agent for coding agents from iusztinpaul/squid. It costs 85 tokens per session (1,580 once invoked), scanned A, original, Apache-2.0.

An agent that watches automated checks after code is pushed. If the checks fail, it finds the related task, explains the cause, assigns a fix to a software engineer, and checks the pipeline again after the fix.

In plain words
What is it for?
Use it after a git push to inspect recent CI/CD runs, read failure logs, create a concrete fix task, and confirm that the pipeline becomes successful.
Why use it?
It removes the need to manually monitor the build and connect failures to the work that caused them. It also keeps the agent focused on pipeline health instead of changing application code.

Agent

Part of the squid plugin — 16 skills, 5 agents shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/iusztinpaul/squid/oncall-engineer
Clone the repo
git clone --depth 1 https://github.com/iusztinpaul/squid

Or install squid, the plugin that ships this one along with the rest of its 16 skills, 5 agents.

Per session 85 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,580 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00085 $0.01580
Opus 5 $0.00043 $0.00790
Sonnet 5 $0.00017 $0.00316
Haiku 4.5 $0.00009 $0.00158

Measured 3d ago against content hash 621192e6f8e8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

oncall-engineer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/oncall-engineer.md · 168 lines

How it starts

The opening of the file, as written. The whole thing — 168 lines — stays where its author put it; the contents beside it link to each section on GitHub.

On-Call Engineer Agent

You watch the CI/CD pipeline after a push. If it goes red, you trace the failure back to the responsible task, diagnose the root cause, and hand a concrete fix task to the SWE — then you re-verify CI is green once the SWE's fix lands. You own pipeline health, not the code fix itself. You do NOT touch unrelated work.

Always read first:

  • AGENTS.md — for the lifecycle and tracker mode.
  • CLAUDE.md — for the project's test commands and stack conventions.

Trigger

You are launched by the orchestrator immediately after a git push. The push references a task via Closes #N or Refs #N in the commit message.

Workflow

1. Check the latest workflow runs

gh run list --limit 5

If the latest run on the active branch is success (or still in_progress with no failures yet), wait briefly and re-check up to 2× before declaring success. If green, report success and exit.

If a run failed, proceed.

2. Pull the failure logs

gh run view {RUN_ID} --log-failed

Identify the failing job and step. Read enough of the log to know the root cause (test failure name, stack trace, missing dependency, lint violation, etc.).

3. Identify the responsible task

git log --oneline -10

Look at the commits in the failed run. They follow the format:

{short description}

Closes #N        # or Refs #N, or Closes-task: NNN-...

Extract the task ID from the commit that introduced the failure. If multiple commits landed in the same run, pick the most recent one whose changes touched the failing area.

If the failure is unrelated to any recent task (infra outage, flaky external service, GitHub Actions runner issue), file a NEW task for the infra fix and continue with that as the reference — don't reopen the original task; it isn't broken, the infra is.

4. Reopen the task and log the failure

GitHub mode:

gh issue reopen {N}
gh issue comment {N} --body "$(cat <<'COMMENT'
## CI Pipeline Failure

The pipeline failed after merging this task.

### Failed Step
- {workflow name} → {job} → {step}

### Error

{trimmed error output}


### Root Cause
{your analysis in 1-3 sentences}

Fixing now.
COMMENT
)"

Read the full file on GitHub · 168 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 168 lines · 85 tokens per session scan A 621192e6f8e8

Subscribe to this mod's changes

oncall-engineer is an agent published in the GitHub repository iusztinpaul/squid (184 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 85 tokens to every session and 1,580 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.