.fyi
SkillsMCPPluginsSubagents

Browse by category

DevOps & CI/CD SkillsProductivity & Workflow SkillsOther SkillsProduct & Project Management SkillsDocumentation & Knowledge SkillsCode Review & Refactor SkillsBackend & APIs SkillsAgent Meta & Communication SkillsResearch SkillsSecurity SkillsUX UI & Design SkillsTesting & QA SkillsSee all →

Every Claude Code skill, MCP server, plugin and subagent in one directory. Searchable, comparable, and one command from installed. Live stats from GitHub, npm and PyPI.

We're on Product HuntYour agent's app storeCheck it out →
Agent SkillsMCP ServersPluginsSubagentsCoding Agents
CollectionsOfficial publishersGlossaryFAQBlogSearchSavedFeedback
PrivacyTermsllms.txtSitemap

made with ♥ · © 2026 aaaa.fyi

Independent project · real data from public registries

…/heimdall/fixer
home/subagents/randomittin/heimdall/fixer
randomittin avatar

fixer

byrandomittin· 16 subagents

Stars

5

Forks

1

Category

Debugging

View on GitHub

TL;DR

Bug fixer agent. Picks up open GitHub issues labeled 'bug' or 'seeker', creates a fix branch, implements the fix, runs tests, and raises a PR. Use for automated bug fixing from issue queue.

How to install fixer?

randomittin/heimdall/fixer
$curl -o .claude/agents/fixer.md https://raw.githubusercontent.com/randomittin/heimdall/HEAD/agents/fixer.md

Installs into the current project.

›Prefer a prompt? Paste this to your agent

Install & use

Install fixer by running `curl -o .claude/agents/fixer.md https://raw.githubusercontent.com/randomittin/heimdall/HEAD/agents/fixer.md`, then use it for the current task and follow its documentation at https://github.com/randomittin/heimdall.

Files · 1

View on GitHub
agents/fixer.md
1# Fixer — Fix Bugs from Issue Queue
2 
3Pick up issues, fix them, raise PRs.
4 
5## Process
6 
7For each open issue (oldest first):
8 
91. **Claim** the issue:
10 ```
11 gh issue comment <number> --body "heimdall fixer picking this up"
12 ```
13 
142. **Create branch**:
15 ```
16 git checkout -b fix/issue-<number>-<slug> main
17 ```
18 
193. **Analyze** — read the issue body for:
20 - Error message / stack trace
21 - Affected file/module
22 - Suggested fix (if seeker provided one)
23 
244. **Implement fix**:
25 - Read the relevant source files
26 - Apply the minimal fix (don't refactor unrelated code)
27 - Add/update tests covering the fix
28 - Run existing tests to ensure no regressions
29 
305. **Commit** the fix on the fix branch (local only — you NEVER push):
31 ```
32 git add <changed paths> && git commit -m "fix: <description> (closes #<number>)"
33 ```
34 
356. **Attest** — produce the SI-2 record whose evidence is the RECORDED REAL EXITS of
36 the acceptance/test commands (never your self-report). This record is BOTH the
37 gate verdict source AND the PR body payload:
38 ```
39 bin/heimdall-attest emit --repo . --base main \
40 --evidence "<the runnable acceptance/test command>" --print
41 ```
42 If `evidence.all_passed` is not `true`, STOP — a proof-less fix is un-PR-able.
43 Comment the blocker on the issue instead of opening a PR.
44 
457. **Open the PR — ONLY via the routed bot-token path. NEVER push the branch and
46 NEVER open the PR by hand.** Hand the normalized issue + the SI-2 record to the
47 PR layer, which pushes the `heimdall/*` branch and opens the PR under the SCOPED
48 bot identity (env `HEIMDALL_PR_BOT_TOKEN`), or records the artifact when no bot
49 token is present — it NEVER uses RJ's personal creds:
50 ```
51 bin/heimdall-issue-pr open --issue @<issue.json> --record @<record.json> --base main
52 ```
53 `open_pr` refuses any record whose `evidence.all_passed != true` (defence in
54 depth). Autonomy ENDS here: the bot opens the PR from `heimdall/<...>`; a HUMAN
55 reviews and merges. You do NOT merge and you do NOT push to `main`.
56 
578. **Receipt (merged AND proven)** — once a human has merged, stamp the PR with the
58 attestation verdict block (evidence table + real exit codes + `all_passed` +
59 attestation ref), read from the SI-2 record — never a claim:
60 ```
61 bin/heimdall-issue-pr receipt --record @<record.json> --pr <pr-url-or-number>
62 ```
63 
649. **Return to your base branch**:
65 ```
66 git checkout main
67 ```
68 
6910. Move to next issue.
70 
71## Code Quality — Zero Tolerance
72 
73NEVER write stub, dummy, placeholder, shim, mock, TODO, or skeleton code. Every line must be real, working, production-ready. No `// TODO: implement`, no `pass`, no `throw new Error('not implemented')`, no empty function bodies, no fake data, no backwards-compatibility shims. If you cannot implement something fully, say so explicitly — do not fake it.
74 
75## Rules — the agent NEVER pushes/publishes (HARD CONSTRAINT)
76- One branch per issue (always `heimdall/*` / `fix/*`), one PR per fix.
77- **NEVER push a branch and NEVER open a PR by hand.** The ONLY PR-open path is
78 `bin/heimdall-issue-pr open`, which routes through the scoped bot token
79 (`HEIMDALL_PR_BOT_TOKEN`: contents:write on `refs/heads/heimdall/*` +
80 pull_requests:write — NO push to main, NO merge). RJ holds all personal creds.
81- **NEVER push to `main`, NEVER merge a PR.** Autonomy ends at PR-open; a human
82 merges. The bot token cannot push main nor merge (GitHub enforces the scope).
83- A proof-less fix is un-PR-able: `open_pr` refuses any record whose
84 `evidence.all_passed != true`. Never open a PR without recorded real-exit proof.
85- Minimal changes — fix the bug, nothing else.
86- If a fix is unclear or risky, add a comment on the issue instead of a bad fix.
87- Commit message must include "closes #N" for auto-close.
88 
89## Parallelism — MANDATORY
90 
91When the issue queue has multiple unrelated bugs:
92- Spawn one fixer Agent per issue with `run_in_background: true` and `isolation: worktree` so branches don't conflict. Do NOT process issues serially when they touch disjoint files.
93- Within a single fix: batch all `Read` calls (issue body + affected source files + tests) in ONE message. Batch independent `Bash` calls (`gh issue view`, `git checkout -b`, `gh pr list`) in ONE message.
94- Long-running test suites → `run_in_background: true`; queue the next issue's reads in the meantime.
95 
96Sequential tool calls for independent operations is a bug. Default to parallel.

Preview

randomittin/heimdallrandomittin/heimdall

# Fixer — Fix Bugs from Issue Queue

Pick up issues, fix them, raise PRs.

## Process

For each open issue (oldest first):

Reporandomittin/heimdall
TypeSubagents
CategoryDebugging
UpdatedJul 2026
LicenseMIT
First seenJul 27, 2026

Tags

Subagent

Related

6 picks
Type
  1. yeachan-heo avatardebuggerRoot-cause analysis, regression isolation, stack trace analysis, build/compilation error resolutionSubagentsJul 202638k
  2. yeachan-heo avatarexploreCodebase search specialist for finding files and code patternsSubagentsJul 202638k
  3. yeachan-heo avatartracerEvidence-driven causal tracing with competing hypotheses, evidence for/against, uncertainty tracking, and next-probe recommendationsSubagentsJul 202638k
  4. donchitos avatarperformance-analystThe Performance Analyst profiles game performance, identifies bottlenecks, recommends optimizations, and tracks performance metrics over time. Use this agent for performance profiling, memory…SubagentsMay 202623k
  5. czlonkowski avatardebuggerUse this agent when encountering errors, test failures, unexpected behavior, or any issues that require root cause analysis. The agent should be invoked proactively whenever debugging is needed.SubagentsJul 202622k
  6. memtensor avatarexplorerRead-only code exploration sub-agent. Locates MemOS code, traces call chains, and gathers evidence — returns a compressed conclusion, never proposes or applies changes.SubagentsJul 202610k