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 skills add EmpiricaAI/empirica --skill forecastgit clone --depth 1 https://github.com/EmpiricaAI/empiricaWrote 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/skills/empiricaai/empirica/forecast)<a href="https://agentmods.dev/skills/empiricaai/empirica/forecast"><img src="https://agentmods.dev/badge/skills/empiricaai/empirica/forecast/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/empiricaai/empirica/forecast"><img src="https://agentmods.dev/badge/skills/empiricaai/empirica/forecast.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00101 | $0.01306 |
| Opus 5.5 | $0.00040 | $0.00522 |
| Sonnet 5.5 | $0.00020 | $0.00261 |
| Haiku 4.5 | $0.00010 | $0.00131 |
Grade A, and why
forecast 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.
How it starts
The opening of the file, as written. The whole thing — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Forecast
Load this when the question is "what comes next", or when a release is about to grade the last forecast.
A direction is a prediction: what will happen, by which release, how sure we are, what that rests on, and what would prove it wrong.
In empirica core, README's What's Coming renders docs/ROADMAP.yaml: the schema, the evidence kinds and their confidence caps are in that
file's header and scripts/release.py enforces them. Read it, do not restate it here. A project without that file keeps its forecast in
its goals; steps 1 to 5 and the gardening apply unchanged, and creating a public roadmap is a question for the human, not a default.
The pass
- Gather (noetic). Open goals (
goals-list), unknowns, open assumptions, recent mistakes, accepted mailbox items (mailbox poll --status accepted), calibration biases, the reproduced-defect backlog, and the last forecast indocs/ROADMAP.yaml. Count with a command. A figure you typed from memory is not evidence. - Draft candidates. Per candidate: one-sentence prediction with a
horizonrelease,evidencekind,confidence(the kind caps it), afalsifiera stranger can observe,impact(low / medium / high), andsteered(is a goal already open for it?). - Sufficiency. A candidate with no measured or reviewed basis is not a direction yet: it becomes a question. So does one whose
falsifier you cannot write as the prediction's own sentence (if you cannot, there are two claims). Check each basis against HEAD:
a report older than the code is not evidence (
git log -S<symbol> --since=<report date> -- <path>). - Ask at most three questions. Rank by impact × (1 − confidence), the answer that would reorder the list most. Each is an
AskUserQuestion(or your harness's ask-the-user tool) with your predicted answer first, marked Recommended, and the one reason that grounds it. If you cannot predict the answer you have not read enough. Attention is the scarce resource: one question that changes the ranking beats five that do not. - Route attention.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2d ago First seen · 71 lines · 101 tokens per session scan A e0aa4b9247b9
forecast is a skill published in the GitHub repository EmpiricaAI/empirica (253 stars, last pushed today), licensed MIT. It adds 101 tokens to every session and 1,306 once invoked, about $0.0004 per session on Opus 5.5. 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 2026-10-26.
Other skills, from other repositories
todoist
Manage Todoist tasks, projects, and labels via official CLI.
asana
Manage Asana tasks, projects, and workspaces via MCP.
request-progress
Group recent conversations into overall requests, review each one's progress from memory, link each to the owner goals it serves, and write the build story, so My Work shows progress meters, shipping velocity and learned working preferences. Use when asked to "group requests", "group by overall request", "review…
owner-goals
Map what the owner wants from the prompts they typed themselves (prompts an agent wrote are left out), then put their top priorities and open asks, each with dated quotes, in the Goals panel at the top of My Work. Use when asked for "owner goals", "my goals", "what do I want", "what am I asking for", "map my prompts"…
agentflow
Orchestrate autonomous AI development pipelines through your Kanban board (Asana, GitHub Projects, Linear). Manages multi-worker Claude Code dispatch, deterministic quality gates, adversarial review, per-task cost tracking, and recoverable pipeline design.
ai-dev-jobs-mcp
Search indexed AI and ML jobs, inspect listings and employers, match roles, and view salary and market stats via AI Dev Jobs MCP.