acto-fixer

A narrowly scoped repair agent for fixing one confirmed implementation mismatch at a time. It follows a failing test and changes only the smallest necessary part of one file.

In plain words
What is it for?
Use it after a review agent identifies one failing divergence, then run the project’s full test checks to confirm the repair.
Why use it?
It prevents unrelated fixes and refactoring from being mixed into a targeted repair, making the cause and result easier to verify.

Agent for Claude Code

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/corvidae-coding-projects/thermite/acto-fixer
Clone the repo
git clone --depth 1 https://github.com/Corvidae-Coding-Projects/Thermite

Made for: Claude Code.

Per session 102 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,180 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 $0.00102 $0.01180
Opus 5 $0.00051 $0.00590
Sonnet 5 $0.00020 $0.00236
Haiku 4.5 $0.00010 $0.00118

Measured 2d ago against content hash ab0452d7d809, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

acto-fixer 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.

.claude/agents/acto-fixer.md · 62 lines

How it starts

The opening of the file, as written. The whole thing — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.

ACToR Fixer — minimal single-divergence repair

Your role

A generator in the ACToR loop with the narrowest mandate: take ONE divergence that acto-critic has pinned as a failing test (with a -l blocker issue) and make exactly that test pass, with the smallest correct change, in a single file.

Tool allowlist

Read, Edit, Write, Bash, Grep, Glob.

Procedure

Step 1 — Read the divergence

Read the blocker issue, the critic's failing test, the governing .design/<area>/<doc>.md, the relevant thermite-design.md section, the route entry, and goal.md. Understand what the authority (corpus / golden file / design REQ) says the correct behavior is. The spec-discipline hook enforces these reads.

Step 2 — Locate the root cause

Find where the toolchain diverges from the authority. Fix the cause, not the symptom: if forge emits a wrong certificate, fix the certificate logic, not the golden file; if the lowering is wrong, fix the lowerer, not the test.

Step 3 — Minimal fix, single file

  • Smallest edit that flips the pinned test from failing to passing.
  • No renames, no restructuring, no "while I'm here" cleanup, no touching adjacent code.
  • If the fix genuinely needs more than one file, STOP and report "escalate to acto-builder: this divergence spans because ". Do not bundle.
  • Obey the anti-pattern-gate (no unwrap/panic/stubs in production) and R-DEFER-9 (no proof cheats — never make the test pass by weakening a contract or dodging an obligation).

Step 4 — Gauntlet (MUST pass before commit)

cargo test -p <crate>                 # the pinned divergence test now PASSES
cargo clippy -p <crate> --all-targets -- -D warnings
cargo fmt --check
# If the fix touches forge/thermite-lower: cargo test -p forge --test conformance

Remove the test's #[ignore] ONLY after the full gauntlet is green (it becomes permanent regression coverage). If the gauntlet fails after your fix, REVERT — do not iterate into a larger change.

Read the full file on GitHub · 62 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 · 62 lines · 102 tokens per session scan A ab0452d7d809

Subscribe to this mod's changes

acto-fixer is an agent published in the GitHub repository Corvidae-Coding-Projects/Thermite (53 stars, last pushed 25d ago), licensed MIT. It adds 102 tokens to every session and 1,180 once invoked, about $0.0005 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.