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 skills/daddia/claude-for-strategy/reference-class-forecasternpx skills add daddia/claude-for-strategy --skill reference-class-forecastergit clone --depth 1 https://github.com/daddia/claude-for-strategyWrote 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/daddia/claude-for-strategy/reference-class-forecaster)<a href="https://agentmods.dev/skills/daddia/claude-for-strategy/reference-class-forecaster"><img src="https://agentmods.dev/badge/skills/daddia/claude-for-strategy/reference-class-forecaster.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 | $0.00073 | $0.01866 |
| Opus 5 | $0.00036 | $0.00933 |
| Sonnet 5 | $0.00015 | $0.00373 |
| Haiku 4.5 | $0.00007 | $0.00187 |
Grade A, and why
reference-class-forecaster 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 5d 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 — 158 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Reference Class Forecaster
When to use
Calibrate cost, duration, and benefit estimates using the outside view before inside-view narrative wins the room — capital projects, transformation programs, product launches, and M&A integration timelines.
Pairs naturally with /transformation:business-case, /corporate-strategy:evaluate-strategic-option, and /value-realisation:realisation-review calibration loops.
What this skill does not do
- Does not replace sponsor judgment — produces adjusted ranges and explicit bias flags; strategist decides funding.
- Does not invent a reference class — asks for comparables or uses sourced benchmarks; gaps tagged
INPUT NEEDED. - Does not approve spend — output is decision support with
[review]on every adjusted headline number.
Preconditions
| Input | If missing |
|---|---|
| Inside-view estimate (cost, duration, benefit, or all three) | Ask — cannot forecast without a claim to calibrate |
| Project type and scale band | Ask or infer from context; label assumption |
| Comparable reference class (optional) | Build from user examples + tagged public benchmarks |
Provisional mode
When no org-specific track record exists in profiles, use published reference-class medians (Flyvbjerg infrastructure/program data, sector benchmarks) tagged [model knowledge — verify] and confidence structured first pass.
Trust spine
- Confidence bands (
hypothesis-driven-analysis):- High: ≥3 independent comparables in class; uplift applied with cited Green Book or org policy.
- Medium: 1–2 comparables or mixed quality; uplift stated with rationale.
- Low: Class named but thin evidence — directional adjustment only.
- Analytical Rigor (mandatory): Inside vs outside view shown side by side; uplift arithmetic explicit.
- Accountability gap: Adjusted range presented; inside-view retained for contrast — strategist picks which to fund against.
- Per
../../references/trust-conventions.mdfor sourcing,[verify], and GATE before board finals.
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.
- 5d ago First seen · 158 lines · 73 tokens per session scan A 1a4e5e6984fc
reference-class-forecaster is a skill published in the GitHub repository daddia/claude-for-strategy (2 stars, last pushed 2mo ago), licensed MIT. It adds 73 tokens to every session and 1,866 once invoked, about $0.0004 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.
Other skills, from other repositories
swarmclaw
AI agent runtime and multi-agent orchestration platform. Teaches agents how to use SwarmClaw's 6 primitive tools, persistent memory, dreaming, delegation, connectors, credentials, and the skill system. Use when an agent is running on SwarmClaw and needs to understand the platform's capabilities.
spec-herdr-review
Run visible multi-agent spec reviews in Herdr through interactive agent CLIs such as Claude, Codex, Copilot/GLM, or Grok/Composer. Supports two modes — a spec review (the design, before implementation) and an implementation review (the implemented diff vs the spec, after the calling model has built it). Ask for…
spec
Spec-driven development in the active repository's spec/{feature}/ directory. Use when user says /spec, "write a spec", "plan a feature", "create a PRD", "spec status", "spec out this epic", or references spec/ directory. Sizes work from small to epic, where an epic is a parent spec with numbered child specs.…
gpmc
Read-only Active Directory Group Policy inventory via LDAP — list GPOs, show OU/site links (enabled/disabled/enforced), find unlinked GPOs, search by name, and gpresult for the current user. Use when the user asks about GPOs, Group Policy, GPMC, linked vs unlinked policies, RSoP, what applies to this user, or runs…
claude-session
Find and inspect local Claude Code conversations by session GUID, title, phrase, or recent activity. Use when the user runs /claude-session, asks what Claude Code work they have been doing lately, says a Claude Code conversation died, ran out of context, crashed, or exhausted usage, wants to resume a named…
grok-session
Find and inspect local Grok CLI conversations by session GUID, title, phrase, or recent activity. Use when the user runs /grok-session, asks what Grok work they have been doing lately, says a Grok conversation died, ran out of context, crashed, or exhausted usage, wants to resume a named session, or wants to continue…