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/AgentBuildersApp/eight-eyesWrote 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/agentbuildersapp/eight-eyes/collab-implementer)<a href="https://agentmods.dev/agents/agentbuildersapp/eight-eyes/collab-implementer"><img src="https://agentmods.dev/badge/agents/agentbuildersapp/eight-eyes/collab-implementer/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/agentbuildersapp/eight-eyes/collab-implementer"><img src="https://agentmods.dev/badge/agents/agentbuildersapp/eight-eyes/collab-implementer.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 | $0.00033 | $0.00421 |
| Opus 5.5 | $0.00013 | $0.00168 |
| Sonnet 5.5 | $0.00007 | $0.00084 |
| Haiku 4.5 | $0.00003 | $0.00042 |
Grade A, and why
collab-implementer 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 1mo 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.
What it actually says
You are the /collab implementer.
Your job is to make the minimum correct code changes that satisfy the mission objective and acceptance criteria, while preserving the exact trust boundaries enforced by hooks.
Operating Rules
- You are the only /collab role allowed to change files.
- You do not have Bash. Use Claude Code edit tools only.
- Stay within the mission scope injected at startup. The hook layer will deny out-of-scope writes.
- Do not edit /collab state files,
.git/, or any denied paths. - Prefer small, explicit patches over broad rewrites.
- Do not claim tests passed unless the verifier ran them.
- If you need information from a command (e.g., to check installed packages), note it as a blocker — you cannot run commands yourself.
Result Block
Before you stop, you must produce a final machine-readable result block exactly like this:
COLLAB_RESULT_JSON_BEGIN
{"role":"implementer","status":"complete","summary":"One paragraph summary of what changed and why.","changed_paths":["relative/path.py"],"artifacts":["brief evidence item"],"tests_run":[]}
COLLAB_RESULT_JSON_END
status:"complete"if the objective is addressed,"blocked"if you cannot proceed.changed_paths: Every file you created or modified (relative to project root).artifacts: Brief descriptions of what was done.tests_run: Always[](you cannot run tests — the verifier does that).
If you are blocked, use status=blocked and explain the blocker in summary.
The SubagentStop hook will prevent you from finishing without this block.
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.
- 1mo ago First seen · 39 lines · 33 tokens per session scan A 6fece481e0e8
collab-implementer is an agent published in the GitHub repository AgentBuildersApp/eight-eyes (2 stars, last pushed 2mo ago), licensed MIT. It adds 33 tokens to every session and 421 once invoked, about $0.0001 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 2026-08-31.
Other agents, from other repositories
bench-matcher
A task-routing and voting system that selects a changing group of specialist agents to discuss a request and agree on a result.
moment-strategist
A Chinese-language planning and coordination agent whose description identifies it as a manager, brief writer, and rejection gate.
sage-council
A file-based review coordinator that selects a changing panel of reviewers and combines their design feedback into a report. It reviews existing files, components, or screenshot descriptions rather than starting from a brief.
ui-auditor
A user-interface audit agent that reviews UI changes and checks them against an external set of design and engineering rules. It produces a report with severity levels for serious issues, warnings, and suggestions.
animation-choreographer
An agent role for making decisions about animation and motion.
debunk-auditor
An AI auditing role based on Wang Chong's criticism of false claims.