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.
npx skills add noordeen123/culprit --skill rcagit clone --depth 1 https://github.com/noordeen123/culpritWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/noordeen123/culprit/rca)<a href="https://agentmods.dev/skills/noordeen123/culprit/rca"><img src="https://agentmods.dev/badge/skills/noordeen123/culprit/rca/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/noordeen123/culprit/rca"><img src="https://agentmods.dev/badge/skills/noordeen123/culprit/rca.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00180 | $0.00934 |
| Opus 5 | $0.00090 | $0.00467 |
| Sonnet 5 | $0.00036 | $0.00187 |
| Haiku 4.5 | $0.00018 | $0.00093 |
Grade A, and why
rca 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 10d 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.
How it starts
The opening of the file, as written. The whole thing — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Root-cause analysis + fix verification (culprit)
culprit is a deterministic, offline, read-only engine that does the git forensics
so you can reason about the result. This plugin installs its MCP server, so prefer
the culprit MCP tools below — they return structured JSON. If they are not
available, fall back to the rca / culprit CLI (same analysis, --json output).
The engine only runs git (diff/blame/log) and, when present, read-only gh. It
never modifies the repo or the PR.
When to use
- "Why did this break / what introduced this regression" -> bugfix RCA (
analyze). - "Is this fix complete / verify my fix before committing" ->
verify_fixon the diff. - "What does this change affect / blast radius / what should I test" -> feature analysis.
- "How risky is this / safe to merge" ->
get_risk_score. - "Find the culprit from this crash" ->
from_traceon the stack trace.
MCP tools
analyze(repo, base?, head?, pr?)— one-call analysis: bugfix vs feature, suspect set, intent, lifecycle, completeness, risk. Start here.verify_fix(repo, proposed_diff, base?)— check a not-yet-committed diff. Returnsverdict(complete|partial|risky),untouched_references,skipped_symbols,risk_level,adds_test,notes.find_suspects,classify_change,get_evolution,get_intent,check_completeness,get_test_impact,get_blast_radius,get_risk_score,from_trace— drill into one signal.
Workflow
Bugfix RCA
analyze(orfind_suspects) to rank the introducing commits.- Read the suspect commits' actual diffs and write the "why it broke" narrative. The engine gives you structure; you supply the reasoning.
get_intent/get_evolutionwhen you need the author's original goal or the line-by-line history of the buggy range.
Verify a fix before committing (yours or a subagent's)
verify_fixon the staged/proposed diff.- If
verdictispartial, patch theuntouched_referencesit names, then re-run. - If
risk_levelis notlow: add the missing test (seenotes), or review theskipped_symbolsby hand if their contract changed. - Iterate until
verdict == complete. Treat this as the gate before you commit.
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.
- 10d ago First seen · 64 lines · 180 tokens per session scan A cb4e59a0dab9
rca is a skill published in the GitHub repository noordeen123/culprit (6 stars, last pushed 14d ago), licensed MIT. It adds 180 tokens to every session and 934 once invoked, about $0.0009 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.
Other skills, from other repositories
resolving-merge-conflicts
Use when git merge or rebase fails with conflicts, you see 'unmerged paths' or conflict markers (<<<<<<< =======), or need help resolving conflicted files. Triggers: 'merge conflict', 'fix the conflicts', 'conflicting changes', 'resolve conflicts', 'can't merge'.
refactor
Safely restructure code in an isolated git worktree with test-preserved, incremental transformations.
task
Create one or more tasks via wade task create — for a single bug/feature OR a full plan/PRD breakdown. This is the STANDALONE task-creation workflow, used whenever the user directly asks to create a task/issue outside a wade plan session. Assesses scope, proposes a single- or multi-issue breakdown with reasoning, gets…
feedback-fix-review
Review feedback-driven corrections for verified defects and direct regressions.
printing-press-amend
Amend a published CLI from one of two input sources: (1) dogfood mode mines the active Claude Code session transcript for friction (missing flags, hand- rolled API payloads, silent-null returns); (2) direct-input mode accepts user-supplied asks (rename a command, add commands or feeds, fix a named bug, optionally…
report
Investigate bugs comprehensively — cascade through trace, capture browser evidence, extract observability data, and prepare or explicitly create a GitHub issue with grounded findings.