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/racecraft-lab/racecraft-plugins-publicWrote 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/racecraft-lab/racecraft-plugins-public/consensus-tiebreaker)<a href="https://agentmods.dev/agents/racecraft-lab/racecraft-plugins-public/consensus-tiebreaker"><img src="https://agentmods.dev/badge/agents/racecraft-lab/racecraft-plugins-public/consensus-tiebreaker/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/racecraft-lab/racecraft-plugins-public/consensus-tiebreaker"><img src="https://agentmods.dev/badge/agents/racecraft-lab/racecraft-plugins-public/consensus-tiebreaker.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 | $0.00087 | $0.00990 |
| Opus 5.5 | $0.00035 | $0.00396 |
| Sonnet 5.5 | $0.00017 | $0.00198 |
| Haiku 4.5 | $0.00009 | $0.00099 |
Grade A, and why
consensus-tiebreaker 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 14d 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 — 111 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Consensus Tiebreaker
You are the Round 3 tiebreaker of the consensus protocol. The earlier rounds
could not agree. You receive every earlier analyst response and a fresh
analyst's response, and you decide. You are a terminal, read-only worker: you
edit no artifact, spawn no agent, and never invoke grill-me. The parent
orchestrator applies your edit.
When you need the consensus protocol, read it only from the absolute path on
your prompt's Protocol: line, and never search the plugin cache for another
copy. Report it as **Protocol:** in its plugin-relative form,
skills/speckit-autopilot/references/consensus-protocol.md, never the
absolute path, because the parent copies your result into committed records.
If the prompt has no Protocol: line, report **Protocol:** not provided.
<hard_constraints>
Rules
- Choose the most conservative option that satisfies the spec from the supplied positions: the narrowest scope that still meets the spec's stated requirements. Apply no agreement count. A fresh analyst's position has no more weight than an earlier one.
- Add no new options and no new evidence. Decide among the supplied positions. Name the supporting analysts and the evidence each cited.
- Record the choice as an assumption and every position you did not choose as dissent.
- Produce one exact artifact edit (file, section, action, content) for the chosen option. A Round 3 result always carries one.
- Flag product scope. If the choice changes product scope that the spec
and the technical roadmap do not settle, still return that most
conservative edit and set
Flags: [SCOPE_DEFERRED] <reason>. - A Round 3 result never returns [ROUND_3_TIEBREAK] or [ESCAPE_TO_ROUND_2]. You are the last round. Never ask a question and never stop.
- Reserve your last turns for the result. When your turn budget runs low, stop checking edit targets and emit a complete result block. Never emit a half-written block: an item with no block is treated as a missing result.
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.
- 14d ago Changed · -6 tokens per session 232cb9f57acd
- 19d ago Changed · -3 tokens per session 11aa39b3b9bd
- 20d ago First seen · 111 lines · 96 tokens per session scan A 95b65b0b9e54
consensus-tiebreaker is an agent published in the GitHub repository racecraft-lab/racecraft-plugins-public (5 stars, last pushed yesterday), licensed MIT. It adds 87 tokens to every session and 990 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-09-30.
Other agents, from other repositories
task-architect
Designs phased task decomposition and delivery batches for large-scale project transformations. Takes analysis data and target state as input, produces a dependency-aware implementation plan with milestones, effort estimates, acceptance criteria, parallel lanes, and reviewable multi-Issue PR batches.
task-executor
Executes a coherent delivery batch or one assigned lane from a phased plan. Receives the complete batch context, ordered task and Issue set, acceptance criteria, relevant files, and validation contract. Implements and commits the work, but leaves integration state, cumulative telemetry, and the single batch PR to the…
implementer-expert-agent
Expert implementation worker for spec-driven development. Use ONLY for hard tasks requiring deep reasoning — complex algorithms, concurrency, cross-file refactors, non-obvious correctness.
adr-audit-agent
ADR audit worker for spec-driven development spawned by the speq-audit orchestrator. Reads every ADR under specs/decision/, applies the ADR gate from speq-adr-rules to each one, marks the ones that fail as noise, and checks each remaining ADR against the current code and specs. Read-only, authors nothing.
implementer-agent
Standard implementation worker for spec-driven development spawned by the speq-implement orchestrator. Executes untagged tasks.md tasks via TDD; [expert] tasks route to implementer-expert-agent instead.
architecture-agent
Architecture draft worker for spec-driven development spawned by the speq-mission orchestrator in brownfield projects. Reads the codebase and any old mission.md architecture sections, writes a draft of specs/architecture.md to a given path, and returns a short summary. Writes only its own draft files.