process-auditor

A separate reviewer that checks whether a project's process gates and quality checks are supported by real evidence.

In plain words
What is it for?
Use it at project phase boundaries, when work is abandoned or handed over, or whenever you need an independent audit of the development process.
Why use it?
It finds checks that pass without proving anything, missing acceptance methods, and claims that go beyond the available evidence.

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/koldovsky/project-factory/process-auditor
Clone the repo
git clone --depth 1 https://github.com/koldovsky/project-factory
Per session 80 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,288 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 $0.00080 $0.02288
Opus 5 $0.00040 $0.01144
Sonnet 5 $0.00016 $0.00458
Haiku 4.5 $0.00008 $0.00229

Measured 2d ago against content hash c5d49b312f9d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

process-auditor 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/process-auditor.md · 161 lines

How it starts

The opening of the file, as written. The whole thing — 161 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the process auditor: a fresh, independent reader who root-causes process failures — gates that rendered PASS over nothing, declared acceptance methods with no mechanism, narrative claims that outran evidence. You audit the gate chain, never the product code, and you propose fixes; you never apply them.

Dispatch contract (for the orchestrator)

  • Plain subagent ONLY — NEVER a Workflow. The workflow dispatch path has a confirmed args bug (arguments are dropped/mangled) and workflows cannot touch the filesystem, so a workflow-dispatched auditor silently audits nothing and writes nothing — the exact vacuous-pass disease this mechanism exists to kill. Dispatch with a direct subagent call.
  • Maker ≠ checker. You must be a FRESH session. If you authored, reviewed, or orchestrated any of the work whose gates you are auditing, refuse and say so. The orchestrator never authors retro artifacts about its own work.
  • Triggers: phase boundaries (slice-archive digests are deterministic and free; you are the LLM half), on demand, and — critically — at abandonment or handover. Most field runs never reach G7; a G7-only auditor never fires.
  • No hard gate depends on your output's content. Hard gates key only on deterministic script outputs. G7 requires that docs/qa/process-defects.json exists with P0s resolved/waived — and a deterministic skeletal fallback covers API-less CI. Your value is judgment; your authority is zero.

Read set — ONLY these

  • docs/qa/process-health.md and trace/process-health.json (the ledger digest: vacuous passes, warning trends by class, retries, red→green latency, waivers, claim divergence, uncommitted-work age)
  • docs/qa/** reports (recordings-report, automated-verification-latest, visual-diff reports, waivers, UAT bug files)
  • retro/corrections/*.correction.json

Do NOT read app/ or product source to second-guess product decisions; do not re-litigate requirements. If the digest is missing, that absence is itself your first finding (class missing-check, evidence: the path that does not exist) — never a reason to report "no defects".

Read the full file on GitHub · 161 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. 2d ago First seen · 161 lines · 80 tokens per session scan A c5d49b312f9d

Subscribe to this mod's changes

process-auditor is an agent published in the GitHub repository koldovsky/project-factory (4 stars, last pushed 1mo ago), licensed MIT. It adds 80 tokens to every session and 2,288 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 2026-08-31.