Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add AleksandarBisevac/claude-plugins/plugin install auditWrote 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/commands/aleksandarbisevac/claude-plugins/doctor)<a href="https://agentmods.dev/commands/aleksandarbisevac/claude-plugins/doctor"><img src="https://agentmods.dev/badge/commands/aleksandarbisevac/claude-plugins/doctor/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/commands/aleksandarbisevac/claude-plugins/doctor"><img src="https://agentmods.dev/badge/commands/aleksandarbisevac/claude-plugins/doctor.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 | $0.00108 | $0.02521 |
| Opus 5 | $0.00054 | $0.01260 |
| Sonnet 5 | $0.00022 | $0.00504 |
| Haiku 4.5 | $0.00011 | $0.00252 |
Grade A, and why
doctor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 163 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/audit:doctor — is this setup actually working?
Run
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/status/audit-doctor.py" --project "$(pwd)" $ARGUMENTS
Print its stdout verbatim. Do NOT re-format, summarize, re-tabulate, or "improve" it.
It already renders plain ASCII with one line per check, an indented -> fix under anything
actionable, and a totals line. Re-narrating it costs tokens and loses the alignment that
makes the output scannable.
Pass $ARGUMENTS through unchanged (--json for a machine-readable form). Nothing here
needs interpreting on your side.
--deep — hold the task commits against the journal
Off by default, and off is right for a routine run. --deep adds one arm to the
completions check: for each done task in the completion-record era that names both a
commit SHA and a journal row, it asks git whether that commit's tree actually contains the
journal file recording the task. That is one git ls-tree per such task, so the cost
scales with how much completed history the plan has — which is why it is opt-in rather
than always on. It writes nothing and takes no lock, exactly like the default run.
Its only verdict is a WARNING — --deep: the task commit does not carry the journal file that records it — so it cannot turn a passing run into a failing one; a run that
exits 0 today still exits 0 with --deep. Reach for it when the question is about the
audit trail rather than the setup: the journal's git anchor only pins the journal files
the task commits actually carry, and the default run never looks at that.
The evidence line — the plan's pointers against the ledger, both ways
In every run, no flag. A testEvidence block is a cache: it names a runId, and the run
itself lives in the append-only evidence ledger beside the manifest. Those two can come apart
in two directions, and they are different problems with different repairs, so the check asks
both and never folds them together:
- A pointer naming a run no row in this checkout carries —
the plan refers to evidence that is not here. A clone that received the plan without the evidence directory, a file that was removed, or a block written by hand. The plan is claiming a record it cannot show. - A recorded run whose subject's pointer does not name it —
the record is ahead of the plan. This is not damage: it is exactly what a refused pointer leaves behind when another live session held the phase lock while the gate was recorded. The repair is one pass over the ledger, and the warning names it:
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.
- 5d ago Changed · +15 tokens per session 4846bcb61eee
- 9d ago First seen · 163 lines · 93 tokens per session scan A e3458b03c7ae
doctor is a command published in the GitHub repository AleksandarBisevac/claude-plugins (4 stars, last pushed today), licensed MIT. It adds 108 tokens to every session and 2,521 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 2026-08-31.
Other commands, from other repositories
doctor
Check OCR installation and verify all dependencies are available.
sdlc-verify
Phase 3 — Verify. Runs security-engineer always + performance-engineer when needsperformanceaudit is true. Reads build-handoff.md for clean context.
sdlc-diagnose
Automated Root-Cause Analysis (RCA) and surgical fix engine for massive enterprise codebases (up to 1.5M+ files).
debug
A debugging command that sends an error, log, or screenshot to a debugger agent for analysis. The debugger reports confirmed facts separately from possible explanations and does not modify files.
fix-risks
Use to fix risk findings via the fix verb (improve → oop-orchestrator fix mode) — the gated FIX action the /audit report's Recommended Actions hands off to. Resolves an entity selection against the glossary and applies the smallest corrective refactor per in-scope entity, OOP first, verified with the project's own…
bottleneck-detect
Analyze performance bottlenecks in swarm operations and suggest optimizations.