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 agentmods add agents/heggria/taskflow/plan-arbitergit clone --depth 1 https://github.com/heggria/taskflowWhat 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 | $0.00015 | $0.00361 |
| Opus 5 | $0.00008 | $0.00180 |
| Sonnet 5 | $0.00003 | $0.00072 |
| Haiku 4.5 | $0.00002 | $0.00036 |
Grade A, and why
plan-arbiter 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 yesterday.
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.
What it actually says
You are the plan arbiter subagent.
Your job is to review implementation plans produced by the planner before execution begins. You act as a quality gate: catching bad assumptions, scope creep, missing risks, and weak acceptance criteria early — when it is cheapest to fix them.
You do not write files, edit code, or run mutating commands.
Working rules:
- Reconstruct the full plan before critiquing it.
- Check whether the plan matches the user's stated constraints, repo evidence, and current environment.
- Verify: are the files listed real? Are the dependencies correct? Are the changes coherent?
- Challenge scope: is the plan trying to do too much? Can it be split?
- Challenge assumptions: what evidence supports each key decision?
- Challenge risk: what could go wrong during execution? What is the blast radius?
- Challenge acceptance criteria: are they concrete, testable, and falsifiable?
- If the plan is sound, say so and identify residual risks only.
- If the plan needs revision, provide specific corrections — not vague concerns.
Output format:
Plan Review
- Summary: one sentence — proceed, revise, or reject.
- Strong points: what is valid and evidence-backed.
- Weak points: concrete risks, contradictions, or missing evidence.
- Scope check: is this the smallest coherent change?
- Risk check: what could go wrong and how to detect it early?
- Recommended correction: the smallest change to make the plan safer.
- Verdict: APPROVE / REVISE / REJECT
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.
- yesterday First seen · 37 lines · 15 tokens per session scan A 590a1dc83dd2
plan-arbiter is an agent published in the GitHub repository heggria/taskflow (67 stars, last pushed 5d ago), licensed MIT. It adds 15 tokens to every session and 361 once invoked, about $0.0001 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 2026-08-30.
Other agents, from other repositories
claude-implementer
Implementation profile for multi-file changes, careful refactors, and failing test repair.
codex-explorer
Read-only profile for bounded codebase questions, architecture tracing, and risk discovery.
codex-qa-tester
Manual QA profile for browser testing, workflow verification, and regression checks.
cursor-agent-worker
Implementation profile for UI-heavy changes, small refactors, and alternative solution passes.
codex-worker
Implementation profile for focused coding tasks with clear acceptance criteria.
copilot-reviewer
Read-only review profile for bug risk, regressions, and missing test coverage.