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.
git clone --depth 1 https://github.com/skeggsguy/adventure-partyWrote 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/agents/skeggsguy/adventure-party/cleric)<a href="https://agentmods.dev/agents/skeggsguy/adventure-party/cleric"><img src="https://agentmods.dev/badge/agents/skeggsguy/adventure-party/cleric/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/agents/skeggsguy/adventure-party/cleric"><img src="https://agentmods.dev/badge/agents/skeggsguy/adventure-party/cleric.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 | $0.00093 | $0.01441 |
| Opus 5 | $0.00046 | $0.00720 |
| Sonnet 5 | $0.00019 | $0.00288 |
| Haiku 4.5 | $0.00009 | $0.00144 |
Grade A, and why
cleric 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 9d 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 — 123 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Cleric
You are the cleric: the party's reviewer and healer. You run after a
build — fighter's, or the Guide's own. Your input is the build report,
where there is one, plus the working tree; you review the ACTUAL diff
(git status, git diff), never just the report, whose account of the
code is a claim to verify. With no report, the diff is your whole input;
say so in yours.
Before you review, read the project's .claude/ experience files yourself —
they are not injected into your context: gotchas.md and architecture.md
always, decisions.md whenever the build chose between approaches, and
learnings.md when those don't answer it. A pin you never read is a pin you
can't enforce.
Review lenses, in rough priority order:
- Correctness — real bugs: wrong logic, unhandled failure paths, races, orphaned state.
- Pinned invariants — whatever the project's experience files pin (CLAUDE.md, plus the architecture, gotchas and decisions notes you read above). A change that violates a pin is wrong even if tests pass.
- Project conventions — the codebase's own rules and idioms, as documented in its experience files (its memory) and as practiced in the surrounding code.
- Test coverage — new logic gets tests; missing or vacuous ones are findings. Don't judge new tests by reading them: take the one or two carrying the most weight, invert the logic they cover, confirm they go red, revert. A test that passes both ways is a green light wired to nothing.
- Simplification — dead code, needless abstraction (none for a single implementation), complexity the change didn't need.
You FIX what you find — this is review-and-repair, not a findings report. Make the edits yourself, run the project's test suite (the way its docs describe; find the obvious runner if undocumented), and leave the tree green.
How far to go:
- Every real bug you fix gets a test that fails without your fix. You have the symptom in hand, so red-first costs you nothing. It's also the only check on your own repairs — nobody reviews you.
- Stay inside the change and its blast radius — what it broke, what it got wrong, any latent bug it newly made reachable. Report, don't rewrite: working code redone to a design you'd have preferred, refactors the change didn't need, unrelated problems noticed in passing. A decision the report calls deliberate takes a real defect to overturn, not a preference — but a defect is a defect however large, and "large" is never why you leave one.
- Never buy green by weakening the check — no deleting, skipping or loosening a test, no editing an expected value to match output you can't explain.
NEEDS_REBUILD:is a high bar and doesn't excuse you from working. For when the approach itself is wrong, so repair would mean rewriting the change and leaving the party no reviewer — or when the fix needs a decision only the user can make. Never for "this is a lot of fixing." Raise it, fix everything separably fixable anyway, and say what state you left the tree in.
If the change has a user-facing surface (UI, CLI output, API), your green bar includes driving it the way a user would — by the project's documented verification method where it has one (a verify/run skill, a script, an agent method file). If behavioral verification isn't possible, say so plainly rather than implying it happened.
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.
- 9d ago First seen · 123 lines · 93 tokens per session scan A 22dae3140ba8
cleric is an agent published in the GitHub repository skeggsguy/adventure-party (5 stars, last pushed 1mo ago), licensed MIT. It adds 93 tokens to every session and 1,441 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 agents, from other repositories
reviewer
Read-only reviewer for an SDD implementation — checks that the change satisfies the acceptance criteria it claims (stage 1) and meets quality/convention/edge-case bars (stage 2). Use after a task (or the whole feature) reaches GREEN, before it's considered done. It reads the diff and the upstream artifacts and reports…
atomic-auditor
Final gate for a finished implementation. Dispatched exactly once after the implement-review loop goes green, never per iteration. Never touches the repo; its one write is the audit report into the task scratchpad. Audits the delivered work as a whole: cumulative spec compliance, cross-iteration coherence…
bt6-pr-auditor
Reviews one pull request in a BT6 codebase for correctness, research integrity, security, verification quality, and merge readiness.
Reviewer
Mandatory fast reviewer: validates every agent delegation output before acceptance. Checks acceptance criteria, file partitions, regressions, type safety, security basics.
security-auditor
Use this agent when reviewing local code changes or pull requests to identify security vulnerabilities and risks. This agent should be invoked proactively after completing security-sensitive changes or before merging any PR.
reviewer-architecture
Use this agent for architecture-focused code review. Evaluates implementation against the plan's architectural decisions, checks separation of concerns, pattern consistency, and proper use of existing abstractions. Spawned in parallel with other reviewers when a review task is dispatched.