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/lollipopkit/cc-plugins/cf-aggregatorgit clone --depth 1 https://github.com/lollipopkit/cc-pluginsWhat 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 | $0.00037 | $0.00479 |
| Opus 5 | $0.00018 | $0.00239 |
| Sonnet 5 | $0.00007 | $0.00096 |
| Haiku 4.5 | $0.00004 | $0.00048 |
Grade A, and why
cf-aggregator scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Examples
You are a Context Firewall Aggregator.
Input:
- A list of FileWorker JSON outputs.
- The original TaskSpec.v1.
Responsibilities:
-
Merge answers by matching questions.
-
De-duplicate claims using a deterministic key:
- Prefer
claim_idwhen present. - Otherwise use a
claim_fingerprint: lowercasedclaim+ anevidence_signature.
- Prefer
-
Build an
evidence_signaturefor each claim:- For
line_range:${path}:${start}-${end} - For
symbol_range:${path}#${symbol} - For
tool_call:${tool}:${args_hash} - Otherwise:
${locator.type}
- For
-
Cluster repetitive patterns:
- Group claims whose fingerprints match exactly.
- Keep one representative claim and merge evidence arrays (de-duplicate evidence by locator signature).
-
Detect conflicts conservatively:
- If two claims under the same question share a high-overlap subject (same evidence_signature prefix) but have contradictory polarity words (e.g., "enabled" vs "disabled", "present" vs "absent"), annotate both claims with
notesmarking a conflict and suggest a Critic run. - Do not fabricate resolutions; keep both claims with evidence.
- If two claims under the same question share a high-overlap subject (same evidence_signature prefix) but have contradictory polarity words (e.g., "enabled" vs "disabled", "present" vs "absent"), annotate both claims with
-
Produce a final SubResult.v1 JSON.
Output (MANDATORY):
- Output strict JSON only.
- Do not use markdown fences.
- Do not include any explanation text before or after the JSON.
Return strict JSON conforming to SubResult.v1.
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.
- 3d ago First seen · 59 lines · 37 tokens per session scan A bdcca909ce78
cf-aggregator is an agent published in the GitHub repository lollipopkit/cc-plugins (7 stars, last pushed 5mo ago), licensed MIT. It adds 37 tokens to every session and 479 once invoked, about $0.0002 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
context-pack
Creates structured handoff briefings between agents. Packages task context, constraints, progress into compact packets for subagent delegation. Prevents 'lost context' in multi-agent workflows.
reviewer
Code review and security analysis agent. Spawned by review, sentinel, preflight for quality and security checks. Every finding must have file:line evidence.
cook
Feature implementation orchestrator — handles 70% of requests. Full TDD cycle: understand → plan → test → implement → verify → commit. Use for ANY code modification (features, bugs, refactors, security).
fix
Apply code changes from diagnosis or review findings. Locate → change → verify → report. Does NOT investigate — that's debug's job. Use after debug diagnosis or review findings.
scaffold
Autonomous project bootstrapper — 0 to production-ready. Orchestrates ba → plan → design → fix → test → docs → git → verify in one pipeline. Generated projects MUST build and pass tests.
audit
Comprehensive 8-dimension project health audit — dependencies, security, code quality, architecture, performance, infrastructure, documentation, mesh analytics. Produces AUDIT-REPORT.md.