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 commands/hex/claude-council/advisegit clone --depth 1 https://github.com/hex/claude-councilWrote 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/commands/hex/claude-council/advise)<a href="https://agentmods.dev/commands/hex/claude-council/advise"><img src="https://agentmods.dev/badge/commands/hex/claude-council/advise.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 | $0.00110 | $0.00960 |
| Opus 5 | $0.00055 | $0.00480 |
| Sonnet 5 | $0.00022 | $0.00192 |
| Haiku 4.5 | $0.00011 | $0.00096 |
Grade A, and why
advise 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 today.
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 — 93 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Ask the council about this conversation.
What makes this different from /ask
/ask sends a question you typed. This sends a slice of the conversation
itself, so the providers can see the reasoning rather than your summary of it.
That is the point: a model given only your framing tends to agree with your
framing, and agreement produced that way measures the framing, not the work.
It is also why every run shows you the payload before anything is sent.
Step 1: Resolve this conversation's transcript
The session id is substituted into this file at invocation:
bash ${CLAUDE_PLUGIN_ROOT}/scripts/session-transcript.sh ${CLAUDE_SESSION_ID}
If that fails, stop and report why. Do not search for a transcript by modification time or pick the newest file: a teammate agent running in the same directory writes its own transcript, and the newest file is frequently theirs.
Step 2: Digest a bounded window
Default to the last 25 turns. --turns=last:N overrides it; --turns=all is
available but say plainly that it sends the whole conversation.
DIGEST=$(mktemp -t council-digest)
bash ${CLAUDE_PLUGIN_ROOT}/scripts/transcript-digest.sh --turns last:25 "<path>" > "$DIGEST"
wc -c "$DIGEST"
Use a plain temporary name. The file path is interpolated into the prompt the providers receive, so a name carrying the session or directory tells them things the digest itself does not.
The digest carries human turns and assistant replies. It excludes tool results, tool inputs, thinking blocks, hook output, teammate and peer messages, and compact summaries.
Step 3: Show the payload, then ask
This step is the privacy control and it is not optional. The script cannot be the gate: inside a subagent the ambient session id names the parent conversation, so nothing the script can read tells it whose words it holds.
Show the user, before sending:
- the byte size and how many turns it covers
- the first few lines of the digest, via
head - which providers will receive it
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.
- today First seen · 93 lines · 110 tokens per session scan A a381b9365831
advise is a command published in the GitHub repository hex/claude-council (688 stars, last pushed yesterday), licensed MIT. It adds 110 tokens to every session and 960 once invoked, about $0.0006 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 2026-09-05.
Other commands, from other repositories
grok-files
List, prune, or gc Grok (xAI) uploaded files. Cleanup for bridge-owned uploads only.
help
Show how to use deliberation here - what each command does, when to reach for it, and real prompts you can paste right now. Read-only.
setup
Configure deliberation with Codex (GPT), Gemini, Grok, and OpenRouter MCP servers.
analyze
Analyze recent runs (latency, tokens, verdict agreement) and suggest model/reasoning/fanout tuning.
ask-all
Ask GPT, Gemini, Grok, and OpenRouter in parallel for independent second opinions.
ask-gemini
Ask Gemini for a single-shot second opinion.