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/allenc84/sapience/loggit clone --depth 1 https://github.com/allenc84/sapienceWhat 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 | $0.00000 | $0.01117 |
| Opus 5 | $0.00000 | $0.00558 |
| Sonnet 5 | $0.00000 | $0.00223 |
| Haiku 4.5 | $0.00000 | $0.00112 |
Grade A, and why
log 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 yesterday.
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 — 105 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/log — Judgment Ledger Command
Usage:
/log <description>— log a new assessment from natural language/log review— show all pending assessments/log review <domain>— show pending for one domain (one of your configuredLEDGER_DOMAINS)/log resolve <id>— score a specific assessment by ID/log calibration <domain>— generate calibration pattern from resolved assessments in a domain (requires 3+ resolved)/log bias— generate cross-domain bias map
Routing
Read $ARGUMENTS and route to the correct behavior below.
NEW ASSESSMENT: /log <description>
If $ARGUMENTS is non-empty and does not start with review, resolve, calibration, or bias:
Extract the following from the natural language description:
- text: The assessment or prediction, stated clearly and specifically. Include prices, names, quantities.
- domain: One of the configured
LEDGER_DOMAINS. Infer from context which domain the assessment belongs to; if none fits cleanly, pick the closest and note the inference in the confirmation. - probability: a number 0-1 for how likely the call is to prove right (e.g.
0.7). Prefer this — it's what makes calibration (Brier score) real. Extract it if the user gives odds/percentages; otherwise fall back to confidence (high/moderate/low, defaultmoderate), which maps to 0.9/0.75/0.6. - horizon: e.g. "3 months", "2 weeks", "end of Q3". Infer from context. Leave blank if not determinable.
- logic: The reasoning behind the call at this moment. Pull from context or ask the user if unclear.
- conditions: Relevant conditions — price levels, team state, market context, etc.
Call mcp__sapience__log_assessment with the extracted fields. Then confirm back:
Logged: [text] | [domain] | [confidence] | horizon: [horizon] | ID: [id]
If the input is ambiguous (domain unclear, no horizon, logic missing), make reasonable inferences and note them in the confirmation rather than asking.
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.
- yesterday First seen · 105 lines · 0 tokens per session scan A 05b765f461ad
log is a command published in the GitHub repository allenc84/sapience (0 stars, last pushed 1mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,117 tokens. 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 commands, from other repositories
weekly
Weekly memory report — facts learned, procedures, repeated mistakes prevented.
btw
Park a stray thought as a sticky without breaking flow — the side-quest capture.
stickies
View and manage Stickies — persistent notes for this project.
recall
Answer from reef memory, reading the pages before answering.
remember
Save something to your reef memory.
whats-new
What the other people's assistants wrote to your coves.