calibrate

A command for archiving several completed OpenSpec changes at once. OpenSpec is a local system for tracking proposed and implemented software changes.

In plain words
What is it for?
Use it to select the correct OpenSpec store, review change status, resolve specification conflicts, and archive multiple completed changes.
Why use it?
It avoids archiving each finished change separately and checks the codebase when specifications conflict with what is actually implemented.

Skill for Claude CodeCodex

Part of the grainulator plugin — 19 skills, 1 agent, 2 hooks, 4 MCP servers shipped together

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 skills/grainulation/grainulator/calibrate
Any agent
npx skills add grainulation/grainulator --skill calibrate
Clone the repo
git clone --depth 1 https://github.com/grainulation/grainulator

Made for: Claude Code, Codex.

Or install grainulator, the plugin that ships this one along with the rest of its 19 skills, 1 agent, 2 hooks, 4 MCP servers.

Per session 16 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 446 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 $0.00446
Opus 5 $0.00008 $0.00223
Sonnet 5 $0.00003 $0.00089
Haiku 4.5 $0.00002 $0.00045

Measured 3d ago against content hash f32fab1470bc, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

calibrate 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 3d 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.

skills/calibrate/SKILL.md · 58 lines

What it actually says

/calibrate -- Score predictions vs outcomes

The user wants to check what actually happened after a sprint's recommendations were implemented.

Arguments

$ARGUMENTS

Expected format: /calibrate --outcome "what happened" or /calibrate <claim_id> "actual result"

Instructions

  1. Parse the outcome: The user provides outcome data as free text or claim-specific results.

  2. Match outcomes to predictions: Use wheat_search to find the original estimate, recommendation, or risk claims that predicted something. Compare prediction to actual outcome.

  3. Create calibration claims as cal### claims with evidence tier production (these are real outcomes):

    • If prediction was accurate: factual claim noting the match
    • If prediction was wrong: factual claim noting the delta (predicted X, actual Y)
    • If prediction was partially right: estimate claim with the refined numbers
  4. Compute accuracy scorecard:

    • Group by evidence tier: what % of stated vs web vs documented vs tested claims were accurate?
    • Group by claim type: are estimates less accurate than factual claims?
    • This validates whether the evidence tier system is predictive
  5. Run wheat_compile.

  6. Print scorecard:

    Calibration results:
    Predictions scored: <N>
    Accurate: <N> (<percent>)
    Partially accurate: <N>
    Wrong: <N>
    
    Accuracy by evidence tier:
      stated: <percent>
      web: <percent>
      documented: <percent>
      tested: <percent>
    
    Next steps:
      /brief              -- recompile with calibrated data
      /research <topic>   -- investigate where predictions went wrong
    
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. 3d ago First seen · 58 lines · 16 tokens per session scan A f32fab1470bc

Subscribe to this mod's changes

calibrate is a skill published in the GitHub repository grainulation/grainulator (86 stars, last pushed 4mo ago), licensed MIT. It adds 16 tokens to every session and 446 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 2026-08-30.