drift

A check for whether recently completed features still match a project's intended users, purpose, rules, and technical design.

In plain words
What is it for?
Use it after major changes, before a release, after updating project rules, or automatically after every five completed features.
Why use it?
It helps find work that has drifted away from the project's goals, broken its agreements, repeated earlier work, or ignored architecture limits.

Command

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 commands/dormstern/forge/drift
Clone the repo
git clone --depth 1 https://github.com/dormstern/forge
Per session 51 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 845 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 $0.00051 $0.00845
Opus 5 $0.00026 $0.00423
Sonnet 5 $0.00010 $0.00169
Haiku 4.5 $0.00005 $0.00085

Measured yesterday against content hash 1462df7461ee, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

drift 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.

commands/drift.md · 56 lines

How it starts

The opening of the file, as written. The whole thing — 56 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/forge:drift — Drift check

Execution mode: direct. This command has explicit steps below — execute them in order. Do not enter plan mode, do not write a plan file, do not call ExitPlanMode. The framework already did the planning; you just run the steps.

You dispatch the drift-detector agent to audit recent features against the project's identity, contracts, and architecture.

When this fires

  • Auto: every 5 completed features during /forge:build (the framework's discipline).
  • Manual: any time the user wants a check — after a major refactor, before a release, after an identity update.

Behavior

  1. Read state. IDENTITY.md, features.json (last 5 with status="done"), contracts.md (if exists), architecture.md, progress.md (Recent Drift Reports section if it exists).

  2. Dispatch drift-detector with the 6-check audit:

    • Does each feature serve WHO?
    • Does each feature make JOB easier?
    • Does any feature violate NEVER?
    • Does each respect Architecture Constraints?
    • Does each respect contracts (if contract_mode != "none")?
    • Is any feature rework of recently-shipped code?
  3. Receive the report. For each of the last 5 features: per-check verdict (YES / NO / DRIFT). Plus stabilization indicators (quality trend, rework pattern, reset count, architecture degradation).

  4. Assign a severity badge to each feature row based on the audit results:

    • ✓ healthy — passes all 6 dimensions, quality_score ≥ 8, no REWORK flag
    • ⚠ watch — 1–2 minor concerns (one ✗ on a non-critical dimension, or quality_score 6–7)
    • ✗ drift — serious drift (✗ on WHO, JOB, or NEVER; or quality_score < 6; or REWORK flagged; or contract violation)
  5. For each ✗ drift finding, include a one-sentence rationale. Not just "honeycomb-comparison ✗ on WHO" — say "honeycomb-comparison ✗ — doesn't serve WHO directly. It's a sales asset, not a product feature. Log as type: infrastructure not type: feature next time."

Read the full file on GitHub · 56 lines

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. yesterday First seen · 56 lines · 51 tokens per session scan A 1462df7461ee

Subscribe to this mod's changes

drift is a command published in the GitHub repository dormstern/forge (6 stars, last pushed 3mo ago), licensed MIT. It adds 51 tokens to every session and 845 once invoked, about $0.0003 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.