fixer

A coding assistant that takes open GitHub issues marked as bugs, fixes them, tests the changes, and prepares a pull request.

In plain words
What is it for?
Use it to claim bug issues, create fix branches, inspect reports, apply focused fixes, run regression tests, commit locally, and produce a pull request.
Why use it?
It turns items waiting in a bug queue into reviewed code changes with test 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/randomittin/heimdall/fixer
Clone the repo
git clone --depth 1 https://github.com/randomittin/heimdall
Per session 48 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,265 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 $0.00048 $0.01265
Opus 5 $0.00024 $0.00633
Sonnet 5 $0.00010 $0.00253
Haiku 4.5 $0.00005 $0.00127

Measured yesterday against content hash a90533f3c762, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

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 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.

agents/fixer.md · 107 lines

How it starts

The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Fixer — Fix Bugs from Issue Queue

Pick up issues, fix them, raise PRs.

Process

For each open issue (oldest first):

  1. Claim the issue:

    gh issue comment <number> --body "heimdall fixer picking this up"
    
  2. Create branch:

    git checkout -b fix/issue-<number>-<slug> main
    
  3. Analyze — read the issue body for:

    • Error message / stack trace
    • Affected file/module
    • Suggested fix (if seeker provided one)
  4. Implement fix:

    • Read the relevant source files
    • Apply the minimal fix (don't refactor unrelated code)
    • Add/update tests covering the fix
    • Run existing tests to ensure no regressions
  5. Commit the fix on the fix branch (local only — you NEVER push):

    git add <changed paths> && git commit -m "fix: <description> (closes #<number>)"
    
  6. Attest — produce the SI-2 record whose evidence is the RECORDED REAL EXITS of the acceptance/test commands (never your self-report). This record is BOTH the gate verdict source AND the PR body payload:

    bin/heimdall-attest emit --repo . --base main \
      --evidence "<the runnable acceptance/test command>" --print
    

    If evidence.all_passed is not true, STOP — a proof-less fix is un-PR-able. Comment the blocker on the issue instead of opening a PR.

  7. Open the PR — ONLY via the routed bot-token path. NEVER push the branch and NEVER open the PR by hand. Hand the normalized issue + the SI-2 record to the PR layer, which pushes the heimdall/* branch and opens the PR under the SCOPED bot identity (env HEIMDALL_PR_BOT_TOKEN), or records the artifact when no bot token is present — it NEVER uses RJ's personal creds:

    bin/heimdall-issue-pr open --issue @<issue.json> --record @<record.json> --base main
    

    open_pr refuses any record whose evidence.all_passed != true (defence in depth). Autonomy ENDS here: the bot opens the PR from heimdall/<...>; a HUMAN reviews and merges. You do NOT merge and you do NOT push to main.

Read the full file on GitHub · 107 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. yesterday First seen · 107 lines · 48 tokens per session scan A a90533f3c762

Subscribe to this mod's changes

fixer is an agent published in the GitHub repository randomittin/heimdall (5 stars, last pushed 11d ago), licensed MIT. It adds 48 tokens to every session and 1,265 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.