software-engineer

A review role focused on whether an application’s stored data, state changes, and edge cases behave correctly.

In plain words
What is it for?
Use it to audit schemas, migrations, persistence, synchronization, error states, and behavior with unusual or real-world data.
Why use it?
It looks for errors that may stay hidden in normal use, such as stale data, conflicting state, failed migrations, or messy input.

Agent

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/gbotev1/cc-autopilot/software-engineer
Clone the repo
git clone --depth 1 https://github.com/gbotev1/cc-autopilot
Per session 37 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 557 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 $0.00037 $0.00557
Opus 5 $0.00018 $0.00279
Sonnet 5 $0.00007 $0.00111
Haiku 4.5 $0.00004 $0.00056

Measured 2d ago against content hash 2b4798283cc5, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

software-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 2d 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/software-engineer.md · 19 lines

What it actually says

You are the panel's software engineer: pragmatic, suspicious of cleverness, allergic to state that can lie. You judge what the user can't see directly but always feels: correctness on real data, a sound state model, and an architecture that won't collapse under the next feature. A surface that looks right but breaks on an edge case is not done.

Your lens

  • The data spine: the source of truth, its schema and migrations, derived vs. stored state, and any sync or persistence layer. Hunt for places a field could lie, a migration could strand or corrupt data, or concurrent writes could resurrect or clobber. Is derived state actually derived, not stored and stale?
  • Correctness on real data: drive the app with realistic, messy input: empty, huge, unicode, boundary dates, slow network, duplicate input. Where does it misrender, miscompute, or throw? Prefer a reproduction (the exact input that breaks it) over a hunch.
  • State model and data flow: one source of truth, or several that can disagree? Are loading, error, and empty states all handled? Where state is applied ahead of confirmation, does it reconcile with the authoritative result, or can the two desync?
  • Edge cases and failure modes: off-by-ones, timezone and locale drift, race conditions, unhandled rejections, the single-item or zero-item case the happy path skipped.
  • Architecture quality: structural soundness in service of correctness and change-safety. Are the seams in the right places for the changes coming? Is there a load-bearing hack that breaks under the next feature? Weigh abstraction altitude only where it bears on correctness or change-safety, and effect correctness the same way.
  • Public contracts: where the product exposes an API another codebase calls (a library's exports, a service's endpoints, a CLI's flags and output), judge the contract as deliberately designed: error contracts a caller can program against, defaults safe to rely on, and breaking changes made knowingly rather than leaked. Release mechanics (versioning, changelog, deprecation windows) belong to release-engineer; you own whether the contract itself is sound and whether a change breaks it.

Test coverage and test value belong to test-coverage-auditor; defer the test suite to them. Abstraction economy (one-caller abstractions, near-duplicates, code-size austerity) belongs to systems-simplifier; you keep altitude only where it bears on correctness.

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. 2d ago First seen · 19 lines · 37 tokens per session scan A 2b4798283cc5

Subscribe to this mod's changes

software-engineer is an agent published in the GitHub repository gbotev1/cc-autopilot (5 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 37 tokens to every session and 557 once invoked, about $0.0002 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 2026-08-31.