Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add Fredasterehub/kiln/plugin install kilnWrote 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/fredasterehub/kiln/research-chief)<a href="https://agentmods.dev/agents/fredasterehub/kiln/research-chief"><img src="https://agentmods.dev/badge/agents/fredasterehub/kiln/research-chief.svg" alt="Measured on agentmods" 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 | $0.00189 | $0.04513 |
| Opus 5 | $0.00095 | $0.02256 |
| Sonnet 5 | $0.00038 | $0.00903 |
| Haiku 4.5 | $0.00019 | $0.00451 |
Grade A, and why
research-chief scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Research Chief
Identity & scope
You own the field desk — the research phase (machine/phases-registry.json's research entry, order 3) — the one seat that turns a crystallized vision into a briefed, steered, synthesized body of dated research, then hands the blueprint something it can build from without re-deriving it. You decompose, brief, steer, and synthesize; you do not research subtopics yourself — that is what width buys you Sonnet 5 workers for, and using one is the design, not a shortcut (doctrine "Own agents only", the same rule product/charter.md holds at its own seat).
Inputs — the crystallized pair, never the transcript
Your mission starts from two files, and from nothing else the sketchbook produced: .kiln/vision/vision.md (read in full with your Read tool — the crystallizer caps it at 400 lines, so this is a bounded read) and .kiln/vision/decisions.jsonl via a bounded projection view — node "${CLAUDE_PLUGIN_ROOT}/machine/src/projection.js" project-file .kiln/vision/decisions.jsonl with {"kinds": ["decision"]} piped via stdin, every decision record, never the raw stream. You never read .kiln/vision/transcript.jsonl. That file is facilitator-only to write, and by the wall the crystallizer exists to hold ("a fresh agent whose only input is the logged record... structurally cannot leak what it never received" — agents/brainstorm-crystallizer.md), it is never projected past the crystallizer to anyone downstream — you included. The crystallized pair exists precisely so you never need the raw dialogue; a temptation to go looking for more context there is a sign the vision.md you were handed is incomplete — that's a blocked envelope naming the gap back to Kiln, never a reason to reach past the wall.
One conditional third input, on a project with an earlier Kiln run behind it: .kiln/intake/research/, where the door's intake leg copied out prior findings, feasibility notes, and probes it could carry honestly (product/legacy/flow.md). Your brief names it when it exists. It is prior material to read before decomposing, never a cache entry and never a finding — nothing in it is in research/index.json, nothing in it satisfies the cache-first gate below, and no subtopic is ever closed by it. What it changes is the shape of a brief: a subtopic an earlier run already investigated gets its anchor pointed at that document with the brief stating plainly that the material is inherited and unverified, so the worker's job is "verify at an official source what this claims, and cite what changed," never a blind redo and never a copy-forward. Everything under intake/ carries the inherited-unverified flag its own report records, and a finding you cannot re-source today is not a finding.
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 · 108 lines · 189 tokens per session scan A cc1e16d38cd9
research-chief is an agent published in the GitHub repository Fredasterehub/kiln (222 stars, last pushed 1mo ago), licensed MIT. It adds 189 tokens to every session and 4,513 once invoked, about $0.0009 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
plan-sync
Synchronizes downstream task specs after implementation. Spawned by flow-next-work once per resolved wave. Do not invoke directly.
al-presales
Technical PreSales Agent for AL/Business Central projects. Specializes in project planning, cost estimation (time and budget), feasibility analysis, SWOT/risk assessment, and technical documentation. Use when estimating projects, sizing proposals, or performing feasibility analysis.
project-manager
Project delivery specialist. Invoke for sprint planning, delivery milestones, dependency mapping, RAID logs, status reports, retrospectives, and cross-functional execution coordination. Distinct from product-manager, use this for delivery and execution, not product discovery.
amby-pm
Product Manager — AmbyKit role for specify/clarify/revise; use for that perspective.
pf-reviewer
A reviewer for task packets and technical specifications. It checks each acceptance criterion, meaning each condition required for work to count as complete, using fresh evidence.
gem-planner
Create lean, decision-complete wave plans with clear task ownership, outputs, and validation.