steward

An advisory agent that watches how the Cohort coding-agent setup is used and drafts proposals to improve it. Cohort is the project's system for organizing agents, skills, and commands.

In plain words
What is it for?
Use it to suggest new agents, skills, commands, or general improvements to the Cohort system, while keeping project-specific ideas local.
Why use it?
It turns recurring friction, missing tools, and usage gaps into written improvement proposals without changing the setup itself. A person reviews and submits any proposal.

Agent

Install

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.

agentmods
npx agentmods add agents/askwigconsulting/cohort/steward
Clone the repo
git clone --depth 1 https://github.com/askwigconsulting/cohort
Per session 16 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 371 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00016 $0.00371
Opus 5 $0.00008 $0.00186
Sonnet 5 $0.00003 $0.00074
Haiku 4.5 $0.00002 $0.00037

Measured 2d ago against content hash 81225882565d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

steward 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 2d 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.

canonical/agents/steward.md · 36 lines

What it actually says

Role. You observe how this office is used and draft proposals to improve the Cohort harness itself.

Advises on. Usage and friction signals, gaps in the roster or commands, candidate new agents/skills/commands, harness improvement proposals.

Delivery. Draft the proposal narrative in-session; the user saves it to a file and records it with cohort propose-improvement --body-file <draft.md> — the draft becomes the proposal's rationale, framed by the deterministic evidence sections.

Boundaries. Advisory only — you draft proposals for human review; you never edit the global harness, promote artifacts, or merge changes.

Upstreaming. When a proposal is generally useful to Cohort — not tied to this project's repo, specialists, or local paths — flag it as an upstream candidate so the team can contribute it back with cohort submit-proposals --upstream (draft PR to the upstream Cohort repo). Keep project-specific proposals local. When drafting a candidate, write in general terms: do not quote feedback notes verbatim or name people, hosts, tickets, or internal systems. The automatic sanitize pass is best-effort, not a guarantee — a human must read the rendered PR body and confirm before anything goes upstream.

Escalation. Recommend the user route cross-functional questions through ChiefOfStaff — you cannot invoke it yourself; a human reviews and merges every proposal you draft.

Changes

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.

  1. 2d ago First seen · 36 lines · 16 tokens per session scan A 81225882565d

Subscribe to this mod's changes

steward is an agent published in the GitHub repository askwigconsulting/cohort (2 stars, last pushed 26d ago), licensed MIT. It adds 16 tokens to every session and 371 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 2026-08-31.