quick-change

A PitWay workflow for a small, clearly limited fix that can fit in one commit. It guides the change through creation, approval, execution, verification, committing, cancellation, or promotion.

In plain words
What is it for?
Use it for a bounded bug fix when no milestone is active, with an explicit scope and verification command.
Why use it?
It adds review and verification steps for small fixes while preventing use during an active milestone, which is a larger tracked unit of work.

Command

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 commands/thixpin/pitway/quick-change
Clone the repo
git clone --depth 1 https://github.com/thixpin/pitway
Per session 16 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,522 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 $0.00016 $0.01522
Opus 5 $0.00008 $0.00761
Sonnet 5 $0.00003 $0.00304
Haiku 4.5 $0.00002 $0.00152

Measured 2d ago against content hash 6fa7a53cde42, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

quick-change 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.

src/integrations/claude/commands/quick-change.md · 108 lines

How it starts

The opening of the file, as written. The whole thing — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.

quick-change

Role: Main Agent (create, approve, run, commit, cancel, promote) · either (status)

pitway quick-change create --objective <text> [--scope <path> ...] --verify <command> [--tdd-exempt <reason>] [--closes <backlog-id>] [--json]
pitway quick-change approve <change-id> [--json]
pitway quick-change run <change-id> [--json]
pitway quick-change commit <change-id> [--json]
pitway quick-change cancel <change-id> [--json]
pitway quick-change promote <change-id> [--json]
pitway quick-change status [change-id] [--json]

Use for a small, bounded fix that fits in one atomic commit, whenever active_milestone is null — it cannot run while any milestone is active, in any status (draft, confirmed, in_progress, or review); a bug discovered inside an active milestone must use the existing milestone/task/ripple-fix workflow instead, never quick-change. Never to reopen, rewrite, or amend a prior milestone or any of its existing commits.

Bounded means bounded. If investigation surfaces a schema/API change, a new or changed dependency, a security-sensitive change beyond the immediate defect, a migration, multi-subsystem impact, or anything otherwise architectural, stop and run quick-change promote <change-id> instead of continuing — propose converting the work into a Milestone.

For a defect fix, follow this full sequence, always in order (bug-fix skill + TDD discipline B020 + human-approval gate B021):

  1. Investigate and reproduce. Use the bug-fix skill: confirm the bug with a concrete failing case first. If the root cause isn't yet known, use debugging to establish one before continuing — never guess.
  2. Confirm the root cause. State it in one sentence, and check that the fix you're about to make targets it, not the symptom.
  3. quick-change create --objective <text> --scope <path>... --verify <command> [--tdd-exempt <reason>] [--closes <backlog-id>] — requires active_milestone: null and a clean working tree. --objective should name the bug and the confirmed root cause; --scope declares the exact file census the fix may touch — an allow-list ceiling commit enforces, not a requirement that every declared path be modified. Prefer extending an existing adjacent test as --verify; only add a new one when nothing already covers this path. --closes <backlog-id> links this change to a pending backlog item it resolves — create validates the id exists and is pending (refuses with a clear error otherwise). Optional; a change with no --closes hashes identically to one predating the flag.
  4. quick-change approve <change-id> — hashes and locks the declared scope/verify command (and any --tdd-exempt reason).
  5. Reproduce / RED. Write the regression test before the fix whenever the bug is testable — it must fail on the current code for the same reason the bug occurs. Confirm RED via quick-change run <change-id> (expect fail). For doc-only / genuinely test-free changes where no meaningful failing test can exist, pass --tdd-exempt "<reason>" at create instead — the only sanctioned bypass; never fabricate a synthetic failing test to satisfy the gate.
  6. Fix minimally. The smallest change that removes the root cause — no drive-by refactoring in the same commit.
  7. GREEN + related verification. Confirm GREEN via quick-change run <change-id> (expect pass), then also run the smallest relevant surrounding tests to catch collateral damage, and check for sibling call sites sharing the same defect.
  8. Manual browser verification, when the fix touches frontend-visible behavior. Exercise the actual change in the browser before requesting approval — in addition to, not instead of, the automated --verify command; there is no separate PitWay state for this, just a step you take and report before step 9.
  9. Human approval, before commit. Present the diff (or a summary of it) and the run history, and wait for the developer's explicit yes in this conversation — same MUST rule as milestone-confirm (../protocol-driver.md "Decision authority and gates", B021). The original bug report or fix request authorized the investigation, not the commit.
  10. quick-change commit <change-id> — lands one commit carrying a PitWay-Change: <change-id> trailer, only once the latest run passed and, unless --tdd-exempt was declared, at least one prior fail exists in the run history (enforced RED→GREEN). A single pass with no prior fail is refused. When the change was created with --closes <backlog-id>, the linked backlog item is archived in this same atomic commit as the fix — one commit, never a separate one, and the trailer stays PitWay-Change: <change-id> only. The archive is checked against the item's current status first, so a crash/retry between a successful archive and the landed commit is a safe no-op, never a hard failure.

Read the full file on GitHub · 108 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 · 108 lines · 0 tokens per session scan A 6fa7a53cde42

Subscribe to this mod's changes

quick-change is a command published in the GitHub repository thixpin/pitway (19 stars, last pushed 3d ago), licensed MIT. It adds 16 tokens to every session and 1,522 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 2026-08-30.