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.
npx agentmods add commands/lugassawan/swe-workbench/audit-codebasegit clone --depth 1 https://github.com/lugassawan/swe-workbenchWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00029 | $0.00933 |
| Opus 5 | $0.00015 | $0.00466 |
| Sonnet 5 | $0.00006 | $0.00187 |
| Haiku 4.5 | $0.00003 | $0.00093 |
Grade A, and why
audit-codebase 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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 59 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Cold-start audit of this codebase across multiple domains.
Step 1 — Parse arguments
From $ARGUMENTS, extract:
--time-box <duration>— default30m. Any trailingm/hsuffix is accepted (e.g.30m,2h).--scope <list>— defaultall. Comma-separated domain names:security,perf,reliability,tooling,testing, orall.--depth <quick|standard|deep>— defaultstandard. Controls fan-out behaviour:quick— single-pass auditor, no fan-out.standard— single-pass auditor, no fan-out.deep— auditor + security-auditor on top-N security findings + debugger on top-N reproducing reliability findings.
- Ticket ref — scan first: any token matching
#\d+,[A-Z]+-\d+, or anatlassian.net/GitHub URL is a ticket ref. Collect first match and remove it from the remaining argument string before further parsing; ignore if none. --top-n <int>— default10. After ticket-ref tokens are consumed, any bare integer in the remaining$ARGUMENTSnot preceded by a flag name is treated as--top-n(integer sniff, same convention ascommands/review.md).
Step 2 — Ticket context (when a ref is present)
If a ticket ref was found in Step 1, invoke swe-workbench:ticket-context with that ref and prepend its summary to the audit context passed in Step 3.
Step 3 — Invoke workflow skill
Pass the parsed values as plain prose to swe-workbench:workflow-codebase-audit:
"Time-box:
<time-box>. Scope:<scope>. Depth:<depth>. Top-N:<top-n>."
Append the ticket-context summary from Step 2 if present.
The skill handles phase orchestration, schema enforcement, fan-out (deep mode), and ranked rendering. This command does not produce Plan-mode output — the audit is read-only by definition.
Output
The swe-workbench:workflow-codebase-audit skill (via the swe-workbench:auditor subagent) produces a ranked findings document. Expect:
- Summary header — scope, depth, and time-box used; total finding count by domain.
- Ranked findings — ordered by severity (Critical → High → Medium → Low), each with: domain tag,
File:Lineanchor, concise issue title, root-cause reasoning chain, and counter-evidence note (what was checked that did NOT confirm the finding). - Domain sections —
security,perf,reliability,tooling,testing(only domains in--scopeare rendered;allrenders all five). - Next-action recommendations — top-N actionable fixes the team should address first, keyed to finding IDs.
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.
- yesterday First seen · 59 lines · 29 tokens per session scan A a84d48888f71
audit-codebase is a command published in the GitHub repository lugassawan/swe-workbench (2 stars, last pushed 2d ago), licensed MIT. It adds 29 tokens to every session and 933 once invoked, about $0.0001 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.
Other commands, from other repositories
dispatch-worker
Run one in-session-agent Worker tick (RFC-0041 §4.3.1).
release-tag
Create and push a version tag after a release PR has been merged.
OPSX: Fast Forward
Create a change and generate all artifacts needed for implementation in one go.
deploy
Build, test, deploy with staged rollout.
simplify
Reduce complexity without changing behavior — code simplification.
gum-schema
Print a JSON description of the active gum command tree, including command paths, aliases, arguments, and flags.